Back to Community
Phasing Out Brokerage Integrations

Hello everyone,

We're shutting down the broker integration feature at the end of September. Everything else on Quantopian will continue unchanged - Quantopian Research, our backtester, our community, and all of our data. We continue to make allocations to selected algorithms, and we're still running the Quantopian Open every month. We've already contacted the few hundred community members who use the feature to let them know. For most of the community, this decision means we can spend more of our time and energy on improving Quantopian. We will be making it easier to learn about quantitative finance, to research strategies, and perhaps get an allocation.

We thought long and hard before we made this decision. A key element of our company philosophy is to keep our goals aligned with our community's goals. This ensures that we're delivering the best experience and product that we can; we're working for you, and for us, all the time. At the end of last year that was still true - we were making allocations and trading our company's capital using the Interactive Brokers integration. Our success depended on the broker integration, and we felt the pain of every limitation and bug in that integration ourselves.

That pain made sure we were investing and improving the integration every day. In April we started trading on behalf of our investment clients using a prime broker, an integration that uses a different codebase. That was the moment when the company goals significantly diverged from the goals of personal traders. Our focus has been on helping the community create high-quality algorithms that we can fund with allocations of millions of dollars, and not on the broker integrations. We have concluded that we can't support personal trading at the level of quality that you deserve and expect.

Moving forward, we want to help people find other ways to trade their own money algorithmically. One possible option is the zipline-live open source project that is derived from our open source backtester, zipline. This project has the potential to be an alternative for personal trading. We've talked to the project leaders, and we've agreed to support the future development of zipline-live. Interactive Brokers has agreed to provide some assistance as well. We support more than a dozen different open source projects already and zipline-live fits our mission. We are also soliciting additional corporate sponsors to help support the project. If you're interested in using this project for your own personal trading, or in helping to build this project please visit http://www.zipline-live.io/.

We launched the Quantopian community in 2012 with just a simple backtester. The feature requests came in fast, and we kept building - first fetcher, and later we added more rigorous alternative data sources. We built broker integrations, history, pipeline, futures, and more. We thank everyone for their help from the bottom of our hearts. The community has pushed us to keep building a high-quality platform. We can't thank you all enough for believing in us and working with us.

I said earlier in this post that our focus has evolved. Still, I want to repeat that our core principles have not and will not change. We're here to democratize finance, to educate people from all over the world, to teach about quantitative finance. We are just sorry that we don't have the resources to do everything we want to. We thank you for all of your support.

Sincerely,

fawce

Disclaimer

The material on this website is provided for informational purposes only and does not constitute an offer to sell, a solicitation to buy, or a recommendation or endorsement for any security or strategy, nor does it constitute an offer to provide investment advisory services by Quantopian. In addition, the material offers no opinion with respect to the suitability of any security or specific investment. No information contained herein should be regarded as a suggestion to engage in or refrain from any investment-related course of action as none of Quantopian nor any of its affiliates is undertaking to provide investment advice, act as an adviser to any plan or entity subject to the Employee Retirement Income Security Act of 1974, as amended, individual retirement account or individual retirement annuity, or give advice in a fiduciary capacity with respect to the materials presented herein. If you are an individual retirement or other investor, contact your financial advisor or other fiduciary unrelated to Quantopian about whether any given investment idea, strategy, product or service described herein may be appropriate for your circumstances. All investments involve risk, including loss of principal. Quantopian makes no guarantees as to the accuracy or completeness of the views expressed in the website. The views are subject to change, and may have become unreliable for various reasons, including changes in market conditions or economic circumstances.

252 responses

Why not charge us for live trading instead of shutting it down?

I would pay

Of course all of us would be willing to pay for the service, and we would be super excited if you actually accepted our offer. The reality is that your business model has changed (now that you have profitable algorithms to sell to investors and all). I plea with you to please give us a further sunset date. 9/29 is way too soon for many of us to learn a whole new system.

I'm disheartened by this as well. Being able to back test on the same platform you're trading on is pretty crucial.

Will this affect over trading at all? Anyway to integrate notifications if buys and sells a stock

I've suspected this would happen sooner or later. Letting people invest their own money algorithmically for free was too good to be true. It was truly a revolutionary feature. Sad to see it go but thanks for the ride.

would pay

I spent a lot of time learning your platform for live trading my algos and you won't let enough time to find an alternative platform. Why not keeping a paid service ?

So much for leveling the playing field. What a massive disappointment.

There are other better tools for running backtests. Other features I'm not so interested in. I came here and invested time into learning your tools just for RH integration. Don't see the need sticking around anymore because this was the only feature that won me over.

at least do you will continue to provide to data (prices and fundamentals) to zipline-live? It must not be free, you could change a resonable fee and I'm sure a lot of us will subscribe the data feed.

I really feel like you are underestimating the impact that this feature had on your customers. For me this was the bread and butter of your business. This was the disruptor, its a shame that you don't continue. Even though there were some problems, it never seemed too bad. I was VERY happy with the service and feel very let down.

I also would pay for this. I have developed a lot here and 30 days for change is unreasonable.

This is they most hypocritical and sad thing I've seen. Let me pay you 5 bucks a month so you can maintain the same features. Is this about supporting the little guys or pleasing the big money? There is no way it is prohibitive to allow us users to stay in this community if we are willing to pay.

One note on backtesting vs a live trade.... Your slippage models are not great. Backtesting is almost deceiving, using this system just to backtest is dumb. I run a live algorithm and i would rather have spent my time elsewhere if i knew Quantopian would drop all moral and ethical obligations when they saw a little cash...

Sold out... real sad... Hope this keeps you up at night in your penthouse guys....

Lashing out will not solve this problem.

Extremely disappointing news. Hundreds of hours spent building algos on your platform and then you pull the plug with a month of notice?

This hurts.

I think it would be better if Quantopian moved to a 'rent-a-server' model, that's what Quantconnect is doing last I checked. No, I'm not associated with Quantconnect. This is going to really hurt the community in my opinion. What else will code written on Quantopian be good for except the contest and Quantopian allocations? If you have an algo that isn't quite good enough to get an allocation, or isn't what they are looking for, but is good enough to trade you'll be out of luck I guess. Betting it all on getting an allocation is a too much of a gamble in my opinion. I'll probably finish up the contest entries I'm working on and move on for the most part. Porting back to C# is going to be a pain. Oh well, C# is my native language....

You are right Jacob, lashing out will not solve the problem. If you identify that the only problem is that they are shutting down the live trading service. Now, if you consider that they are hardly giving any notice before shutting it down as a problem, that is worth lashing out over. Many of us make secondary income running our algorithms, and ~40 days is simply not enough time to shift to a different service (if one even exists).

If you run a long only strategy with RH... you are completely out on your own... Quantopian already has the RH API developed as they were granted access.

Regarding using IB...Pipeline isn't incorporated yet... I don't know anyone that runs an algo that doesn't incorporate pipeline filtering... clocks ticking... Im sorry to "lash out". But this is the time to stand up for values. What is happening here is not about resources. Its nieve to think that.. this thread has over 300 views and most people would be willing to pay 10-30$ a month to keep in this game...This is how the rich stay rich. see life for what it is.

Removing live trading will KILL this community. People aren't going to devote hundreds of hours to be able to execute a great backtest. People get into this out of passion, they stay in because they get paid for following their passion.

Moving to a paid service would provide a passive income stream for Quantopian - And enable us to continue benefiting from the passive income stream Quantopian has allowed us to create for ourselves.

People aren't going to continue to write algorithms with the hopes that you will give them an allocation - Anyone who can get an allocation doesn't need quantopian when they weigh the risk to the reward (spend hours upon hours writing an algorithm that most likely won't be accepted, and, if it is, will yield a very small profit given the current allocation sizes and commissions)

@Warren QuantConnect has full python support these days. Seems like a solid alternative.

Oh well... back to trading manually until I find something else to migrate to!

@jacob you are correct. I spent a hundred or so hours developing what I've done because of the promise on actually integrating it into the market. Back testing here and then moving the logic to another platform simply doesn't work.

What is the point of Quantopian if you can't execute live? I'm new to the service, and it makes no logical sense to learn the system only to have to translate the code to another source. Sounds like this company is collapsing.

Live trading on the same backtester platform is one of the major draws of Quantopian. The majority of sophisticated traders will be less motivated to develop on the platform if there is no reward of being able to actually trade what you just created. We could pay for the service, but I don't think Quantopian is fully focused on enhancing the platform and trading experience. Instead they are more focused on cultivating the hedge fund and recruitment. Paying for a second rate service would not be the ideal scenario. Additionally, Q may find out that recruitment may take a hit without live trading.

Quantopian is still a good backtesting tool and the Pipeline API is extremely powerful, but the live trading was always unstable. This may be a good opportunity for people to start experimenting with the IB API and hosting their own live algorithms or experimenting with other platforms such as Quantconnect, which is more focused on the technology. Feel free to reach out if you are looking to do the same.

Will the paper trading with Quantopian still work?

I recommend you all go leave a review on their Facebook page as well.

I recommend you do something more productive than leaving a review on a page that very few people have ever visited/will visit. Instead of acting on your emotions, act on logic and utility. Spend that time solving whatever problems that this has caused for you. Complaining will not solve your problems - it will waste your time.

It is unfortunate to see this go. I mentioned this before, but I think the live-trading element was an incredibly clever loss-leader for Quantopian to attract and grow their userbase. It's what brought me here.

There are some really intelligent and knowledgeable professors of economics and those types participating in the Q Open. I on the other hand don't come from that background. So I gotta ask myself, is it worth it spending months on end researching and exploring potential alpha sources on the Q platform when my only chance of getting anything out of it means I need to beat experts in the field (in addition to a whole lot of luck)? The value proposition for developing on the Q platform is a bit more bleak without the live-trading element. Put another way, with this change I'm more likely to make money developing my code on another platform where I can live-trade it than I have any chance of getting anything out of Q Open or Q Fund. Those efforts are no longer unified.

On the other hand, once upon a time I was the CTO of a venture-funded start-up, so I know all about the importance/pressure to keep a singular focus and not spread yourself too thin. I could see this going either way -- I guess time will tell whether Quantopian just shot themselves in the foot.

I haven't used Quantiacs, but what does Quantopian have now to differentiate themselves from Quantiacs? With this change it would appear Quantopian has less of a moat, less value proposition over Quantiacs.

So, y'know, I will keep on working on ideas for the Q Open and Q Fund -- the research environment here is amazing -- but now my efforts will be divided instead of unified between two platforms, with putting food on the table (so to speak, via live-trading) necessitating more emphasis -- for now.

Thoughts:

  • It would be amazing if we could get access to that prime brokerage through the Quantopian platform. :) Problem solved, huh?

  • There is another platform that supports live-trading: Quantconnect, but they appear to be running on computers from the 80s... ran out of memory trying to backtest their demo "Buy-and-hold SPY" algorithm. Also, they don't support Robinhood.

  • I'm sitting here on an computer with 64gb ram, 6 cpu cores, and a recent graphics card. It's been kind of comical developing on a super slow cloud platform. I'd love to be running my backtests and live trading locally. But....

  • There's the zipline-live project. They don't have fundamentals or pipeline or Robinhood integration. Briefly glossing over it I didn't even see any mention of data.history(). In addition they warned me it would be moderately difficult to get it up and running. Not encouraging so far...

  • It's not hard to build a live-trading platform. Problem is the data -- it's so massive and expensive $$$$$. Perhaps there'd be enough community interest to build out a "quant collective" platform that solves these issues?

Fawce, Very disappointed to see this.

I do believe the real trading integration was a major driver for joining this platform and wanting to get the algorithms to actually work with the intent of allocating ones funds to it. I have invested over 18 months of work in this platform and although I was concerned about the future commitment to this feature I felt your mission for wanting to maintain a level playing field for the smaller guys would win out. This move significantly reduces the value of using this platform specially now we know that at any time you guys can decide to take something else out. Who knows ...

The end of life notice is too short and not respectful of the time of the community members who have to actually figure something else out. Please at least reconsider that while we all figure out what to do !!!

If it buys us 90 days instead of 40 days, it's not time wasted. -and they do have a decent following on there.

This is truly unfortunate though. I guess, the service was free to begin with so who am I to complain right? Regardless, I know a lot of people put hundreds of hours into their algorithms including myself. We're now faced with porting to another site to risk them eventually doing the same thing and ending service, or creating your own platform which would take hundreds of additional hours. You gave some kids some really cool toys that opened their mind to all possibilities and opportunities of the world, only to take it away after they've seen a glimpse of what they can't have.

Noooooooooooooo. I am just starting...

This is disappointing. I can't imagine that maintaining an API link to IB be that difficult or expensive. It is working now and only requires some debugging to move forward (I can only guess). This effort is most probably minuscule compared to the effort of the community to migrate to another platform.

It is working now

That is a stretch. Not sure I would necessarily call it working.

If the issues were actually about maintaining the API and not a conflict of interest with their new rich friends then i would assume that they would plan for the full development/porting of zipline live/piepline and all the promises on that web page to github before announcing a cut off date... Just as a gesture of good faith. Seems like there isn't any good faith in this rushed withdraw from their values and mission statement.

I would happily pay a reasonnable monthly fee (10s of $ not 100s)

I wonder if not allowing auto trading is going to drive some people off the site that would otherwise also compete and thus lower the level of the competition or are those different categories of people? I for one was more interested in my own trading but I was on the side developping multistrategy algo that I was hoping to present in the contest at some point...
However the prospect of winning any material money in the contest is not the driving force as it remains tough to win as far as I can tell... now I ifnd myself googling for alternative to Q for my own trading... turns out there are not too many good solution out there!
QC is a pain to develop with, you can sorta use collective2 api but its pricey for little added value, I have not tried quantiacs or couldtrader , has anyone good alternative to suggest for my home-grown algos?

There is no doubt that this is a devastating loss. There aren't (that I am aware) any other places which provided enough cross-sectional (fundamental) data to do stock screens, integrated seamlessly into both a freeform research environment, algo paper trading, and live trading. Although I was forced to shut off my algos for work last November, I benefited greatly from the workflow, and I will miss it.

That said, this should come as a surprise to very few people. It was clear that despite all their nudging, many people who were live trading their own accounts were not trading the sorts of algos that Quantopian was interested in financing. In their business model, that makes them strictly parasitic. Perhaps they paid their way in terms of contribution to the "community", or perhaps not. I suspect that the forums/community add little in the way of concrete long-short algos which are suitable for their fund. In fact, I suspect that many of the people who are able and willing to write algos which cannot be traded personally yet fit the mold for an allocation do so quietly and with little interaction in public.

I hope people have some sympathy for how difficult a decision this must have been for them. At the end of the day, they have to keep the lights on, and they have employees whose livelihoods depend on the business being a success. Personally I think the "Quantopian execs have turned into plutocrats in penthouses" angle is totally wrong. They are running a business in a tough year for the industry, and just trying to move forward with a focused plan. Live IB trading sounds like it's been a bit of a mess for the last few months, and probably quite a stressful distraction.

Still a terrible shame, though. Especially the data... I was never much a fan of the Python...

Very disappointing news... It will make no sense to spend hours in this community anymore if the perspective to run your own algo will not be there anymore! Very sad. Looks like it is dying... That is what big money does - it is hard to conciliate open source, community driven development, and venture capitalists.

I am still a bit confused. Can you still do live paper trading on Quantopian's platform?

I agree with you Simon, However if your mission statement is " Level wall streets playing field" and then you sell out to wall street...

I think that Mike Dent, Dan Dunn, John Fawcett, Jean Bredeche and the rest of the board owe us an explanation.

Thats some deceitful stuff right there. I developed a long only RH algorithm... but I'm in the same boat i understand that hedging and any market is a better play against risk. Eventually i will adapt my method to short on similar principals. However they only way that would ever re-enter into qunatopians acceptable competition algorithms is if i can continue to develop on a very similar platform.

I am sad, as I trade real money, and haven't got a plan B for someone who can do it, even for a simple strategy. However, running my own company, I realise that commercial realism prevails, and they would have agonised over this one.

@Kevin - Live paper trading is not being axed, from my read, just the broker integration.

I don't need anything other than daily price data and stuff I can get from Quandl, so probably a hosted version of zipline would suffice. Then I would not need to recode. I agree with the comments that $10s but not $100s of dollars per month for the platform is comfortable.

Even at $150/mo it is nothing vs fund mega millions, there aren't a huge number doing live trading, although a subscription would reduce the volume level over algos focused on only 1 or a few stocks. 78 messages talking XIV, VXX etc in August alone come to mind. I can't know how many were just here using a free service without understanding the deal or were (or would become) also interested in the fund or contest, maybe even most, I can't know, but that type of thing was viewed as a distraction. Also the login issue complaints that were appearing unresolvable disappear. I wonder how many understand the program or sort of view it as a birthright utility.

Since Jan 2014, I lost some money on IB, have an RH algo that has tripled in 9 months and is running ~75% or so of parallel paper trading. 4M in ~5 years in a backtest so it looked great, had hopes for my future in it. 2678 algos stored locally (many are modifications of each other), major investment of life, time. My RH code was the best return so far even tho the time invested was relatively tiny, for example ...

Today 17 tabs open running variations of code toward potential contest entries or fund candidates (just finding it really tough at 10M), so even though the fund has been my main focus, on this sudden loss of real money trading, can't process it, except maybe with rough thoughts like wondering whether freeloaders ruined some things, I don't want to characterize them that way, I'd like to think all users arrive with an innate sense of fair exchange, reality, yeah maybe not so much or I don't know.

For some slower strategies (monthly or quarterly), one can still do it manually by live paper trading I assume? So not a total loss considering we'll still have access to a lot of fundamental data.

And I do see why they may have wanted or needed to make this move. But I would have thought a migration to github for a lot of the code that they don't need anymore would have been a nice gesture, and one that wouldn't have taken too much time or support?

Still though... I think they missed a key element, many of us came to this platform to use live trading first, then we started experimenting with developing market neutral strategies. It wasn't the other way around. So by now discontinuing that service they will shut off a key driver of user growth and therefore of viable algorithm development growth.

Simon, I agree with your statement. I would self-identify as a parasite to their business model to a degree. However, I did try to contribute to the forums and answer questions when I had time. It's unlikely, but possible, that my contributions aided someone that wrote a contest winner, or a helped a future contest winner. Further, I was only a parasite to the point that I still had a lot to learn to develop a strategy, and write an algorithm that would be worthy of, and eligible to enter into the contest, but I had a desire to do so. That being said, they are providing extremely short notice at the expense of those who helped build their business.

also note that IB doesn't even integrate with Turbotax. like now i need to find a tax accountant to do that too or generate 1000000's of pages of PDF and submit them online through their CSV generator thing....

This is just classic wall street, evil big money fk the little man wall street. I came here cus of the community. Turns out they wanted to fleece a few of our ideas and leave us out to dry after group think was over... Wouldn't be surprised if they start steeling our code content or looking up high returns and copying code at this point... Did you read the terms of service?

well this is bullshit, we spent month working on our algorithm

It is amazing how people can complain about a free service. Grow up.

'Tis better to have loved and lost than never to have loved at all.

Definitely a bummer. Might have to reconsider how badly I actually wanna learn this stuff considering the fact I'm definitely not gonna win a contest any time soon.

From Jared Board (QuantConnect Owner)

As a guesture to welcome all Quantopian users the new baseline free
tier is 8GB Ram allocation! Enjoy!

@jacob do we need a special invite? I'm joining QuantConnect

I would love to be able to read between the lines of this announcement. There might be different reasons for this decision. Quantopian knows this is going to be a though hit to the community and if they are willing to accept it either Quantopian business is going so well they can focus solely on what they need without caring about the community (I doubt it) or, more realistically, they have limited resources and they have to focus only on what is essential to their business because they are facing dire straits. If so, I wonder if the next step is Quantopian shutting down. I know there have been so many announcements about $250M of Investment and Multi-Million Dollar Allocations so that everything seems fine, but you never know what really happens behind the door.

@luca: we can speculate till we are blue in the face - it does us no good.

Jacob,

are you quite literally the machine.

Free services value customers all the time...

I agree with the gist of the comments made here:

1) there is no reason Q should support live trading algorithms that don't contribute to Q's bottom line for free. Startup businesses are tough and burn money fast. Q is just as susceptible to this as the next outfit, and it is completely understandable that they must operate in a cash flow positive way.

  • however why not offer a subscription fee that includes the cost of supporting logins, code changes, etc?
    By doing so Q is amplifying its reach into the world of quants, and possibly involving that rare programmer who may contribute to their best
    algo solutions. Moreover, Q is thereby remaining much more VISIBLE to quants and financeers worldwide. For example, while laboring to develop algo worthy of winning one of the prizes, I was always amenable to the idea of running real money through a proven Q app and paying the 2%/20% fees inherent in that. Granted I am a small player, managing a measly $7 million in capital, but multiplay that by several 100 other RIA's lurking in the background, and you are talking real money.

    Of course I can keep programming away for the sake of the simulated trading environment. Except that I would be hesitant to do that, as we know how different real live trading can be from simulations.

    I cannot imagine that that would take more than the efforts of 1 full time technical person at Q. Assuming a salary of 80k a year, and dividing that
    by 5000 plus users, it seems the cost of that would be relatively trivial. Add to that of course the costs of the data feed. How about it Q? What is a reasonable cost for such a monthly service?
    to permit (e.g extra 30 days seems reasonable) migrations to other platforms / languages that
    permit migration to other interfaces/solutions ks to IB and Robinhood. Thousands of programmers have contributed months, and in my case years of their programming time helping Q members achieve this record-setting trading environment. They should not be hung out to dry.

    I trust the Q team is not out to harm the thousands of programmers who believed in their model and gave it their all. So I am anxious to hear in the next few days what path to a better transition they are able to offer that provides a "win/win" for all concerned.

@jacob shrum I am speculating because I really like Quantopian. The choices they made in developing their platform make happy both the software engineer and the scientist that are in me. It's hard to find something similar. So I'd really like to know if the next step is closing down Quantopian.

@luca, they will not publicly admit that they are failing - so you won't get an answer. I understand your passion, but only time will reveal the answers you are looking for.

@jacob, you are right, but at the same time I like to hear what other people think. It doesn't change much, I still have to make my own choices while I wait for the answers but it's interesting to hear other point of views.

@Luca I suspect they are NOT going under, given they raised $25m in November:

https://www.crunchbase.com/organization/quantopian#/entity

https://www.sec.gov/Archives/edgar/data/1589635/000123191916000067/xslFormDX01/primary_doc.xml

From Andreessen Horowitz. They are top tier. Basically Quantopian has won.

In fact, why not, here's a quote from Ben Horowitz himself, from his book (a must read if you run your own startup):

“Every time I read a management or self-help book, I find myself saying, “That’s fine, but that wasn’t really the hard thing about the situation.” The hard thing isn’t setting a big, hairy, audacious goal. The hard thing is laying people off when you miss the big goal. The hard thing isn’t hiring great people. The hard thing is when those “great people” develop a sense of entitlement and start demanding unreasonable things. The hard thing isn’t setting up an organizational chart. The hard thing is getting people to communicate within the organization that you just designed. The hard thing isn’t dreaming big. The hard thing is waking up in the middle of the night in a cold sweat when the dream turns into a nightmare.” ― Ben Horowitz, The Hard Thing About Hard Things: Building a Business When There Are No Easy Answers

The main thing I'm worried about with quantopian is thus: How do they expect to draw enough users to keep coming up with good algorithms for their investors when there is almost 0 incentive to do so?

Prior to this, you had a crowd of people live trading and backtesting in order to live trade, sharing their knowledge and strategies which helped the people making algo's for the fund. Now you will only have contest algorithms, but barely any users left to collaborate with and draw from. I've been at this a year and I know there is hardly any chance of getting ranked in the contest or getting picked for an allocation due to their very institutionally driven rules so why would I even bother attempting? Why would anyone other than a few ultra-knowledgable and ultra-dedicated people even attempt it? Surely quantopian as a fund will suffer.

So could someone summarize quant connect real quick? Want to know if it's worth switching over.

Jacob, You really think they are failing to make ends meat? like wouldn't a subscription base allow them a solid market free capital stream? additionally rationally thinking the highest i would pay a month is likely 30-40$ Something like that would reduce the noise and logistical concerns.

Regarding the original posting " We will be making it easier to learn about quantitative finance, to research strategies, and perhaps get an allocation" - J.Fawcett - I spent 6 months developing something... When i started live trading i needed to re-learn everything about the market and do it fast cut i was loosing money. If you remove this feature the quality of your base will change.

@John Fawcett - Charge me homie.

I will only say this to all the other users with me: lets jump start zipline live project and make it what quantopian was, is, and should be. It's ALMOST READY. we can get it going in a month and not lose any progress!

Realizing how difficult it is to replicate what Quantopian has been offering us for free... all that data... newfound appreciation for it. :)

I have a suggestion for people who are panicking. If your algo isn't terribly time-sensitive (which I assume most here aren't), why not just paper-trade your algo and use something like Monkeyscript to scrape the trades as they happen 15-minutes delayed (perhaps you can even compensate for it), and execute those trades via whatever brokerage API endpoints you wish to use?

And just like that, once again, the majority is left holding the bag. Quantopian used us to get to where they need to be. Lesson learned!

I don't understand why Quantopian wouldn't have looked to sell or partner with another company to use their live trading feature and support it? Namely, wouldn't Robinhood find this aligns well with their business model? Maybe Quantopian is hoping to get us riled up and therefore use that as a bargaining chip to help sell something they've already developed? Maybe QuantConnect would be interested in purchasing this and/or integrating it into their platform?

If you do the math, there are probably somewhere on the order of 10,000 users that would be interested in paying $200 a year? That's $2M a year of revenue that could be paying out right now. And presumably the base of users has been growing and would continue to grow.

1) I would pay an asset based fee (tiered, or flat + tiered) but the 1 million dollar restriction has to go.
2) Only a 5 week heads up is kind of rough...

Seems like providing the platform as a [paid] service is a bigger market opportunity than crowdsourcing alpha models for a hedge fund. Have you considered spinning this out as a separate business?

@Burrito Dan. I like your quote, but what is your view then? Isn't this decision affecting its user base and so hurting its business? If Quantopian doesn't have financial problem, why taking the risk of ruining their business with this drastic decision? I am asking for the sake of entertainment ;)

@Thomas Havens - The Robinhood API is not officially public, but it's also no secret. It's the same endpoints that the app uses. Just google it -- it's really easy to use.

@Luca OK then:

“IF YOU ARE GOING TO EAT SHIT, DON’T NIBBLE” ― Ben Horowitz, The Hard Thing About Hard Things: Building a Business When There Are No Easy Answers

Might be time for Quantopian to update their community tagline.

Our community — 100,000 members and growing — ranges from seasoned algorithmic traders to aspiring quants. We help each other with code problems and discuss ideas in algorithmic trading.

I expect a huge majority of the community is going to move on to other platforms that allow live trading, seeing as how, at least for me, the live trading ease-of-use was the solitary reason for using this platform.

Support the move - I think institutional clients would be more comfortable too that the code they essentially allocate to can't be easily used to fund another account. The more clients the better experience for users like me :)

Thank you all for replying. Your passion shines through, and I appreciate it.

Several of the replies we got today asked about whether it would help if we started charging for the service. Unfortunately, that wouldn’t solve the problem. One of the big reasons we’re making this choice is so that we can focus on our business: helping thousands of people each month learn about quantitative trading, teaching them about statistics and signals and coding, and rewarding the best of their ideas with prizes and allocations. We can do that more, and better, if we improve our focus.

Some people ventured estimates of the size of the personal trading business. We only have a few hundred people trading real money. If we started charging $20/month (suggested earlier in the thread), and every single person agreed to pay, it would only be a fraction of the total cost. I just shared a longer version of this answer here.

Several replies were about alternatives, and there were several reasonable ideas kicked around. I suggest looking at http://www.zipline-live.io/. Here at Quantopian we are making continuous investments in zipline, and zipline-live can naturally piggy-back on those improvements. We have seen the power of open source at Quantopian. I am hopeful that the time and energy that people have poured into their algorithms can spill over into zipline-live. With your excitement and skill, zipline-live might be the replacement that so many people here are looking for. The maintainers of zipline-live can better answer questions about how to integrate data sources, add pipeline, and important details like those.

A few people asked about what Quantopian does other than brokerage integrations. Quantopian is the biggest community of quants around, and most people come here to learn. We provide tutorials, lectures, tools, and tons of data. For some people who use Quantopian, it’s about working on an idea that might someday get an allocation. I write this paragraph to extend perspective, not as a solution.

I want to close by repeating fawce’s thanks from before. You’ve all pushed us to make Quantopian a great platform, and you’re still pushing us. Thank you, all, very much. Ending this feature is very hard for us. Many of you had kind words for us in this thread, and understood how hard this is for us, and I thank you for that empathy. For many, we know we’re disappointing you, and we are very sorry.

Disclaimer

The material on this website is provided for informational purposes only and does not constitute an offer to sell, a solicitation to buy, or a recommendation or endorsement for any security or strategy, nor does it constitute an offer to provide investment advisory services by Quantopian. In addition, the material offers no opinion with respect to the suitability of any security or specific investment. No information contained herein should be regarded as a suggestion to engage in or refrain from any investment-related course of action as none of Quantopian nor any of its affiliates is undertaking to provide investment advice, act as an adviser to any plan or entity subject to the Employee Retirement Income Security Act of 1974, as amended, individual retirement account or individual retirement annuity, or give advice in a fiduciary capacity with respect to the materials presented herein. If you are an individual retirement or other investor, contact your financial advisor or other fiduciary unrelated to Quantopian about whether any given investment idea, strategy, product or service described herein may be appropriate for your circumstances. All investments involve risk, including loss of principal. Quantopian makes no guarantees as to the accuracy or completeness of the views expressed in the website. The views are subject to change, and may have become unreliable for various reasons, including changes in market conditions or economic circumstances.

@delman they have plenty of data on their users behaviour, and I'm sure those of us who live trade are in a minority they can afford to lose

@Dan Dunn

Thanks a lot for the reply and confirming low numbers of successful traders. i assume that also means that algorithm allocation isn't actually impossible...

1) When will pipeline and the equivalent of the bare bones Quantopian be online?
2) Robinhood API needs to be part of this build as the code is out there and robhinhood is much better for learning than IB. ( Tax integration, free, app ect.... hedging is an important tool but over mosts heads lets be honest)
3) if you are truly dedicated to the learning of this filed you must understand that working for self reward is the primary motivation... This does directly contradict your statement about wanting to be a learning platform. Khan academy statistics is more than anyone really needs to know . You don't need to re-teach that stuff. your value is in your platform.

@All,

I am saddened by this news. I am going to assume that this decision is final and there's no going back.

With that said, if anyone has interest in migrating your Python codes to TradeStation or MultiCharts, please email me at: [email protected]

I highly suggest checking out EasyLanguage for TradeStation and MultiCharts. You can do a lot of the things you do here on Quantopian with EasyLanguage and it is very easy to learn (obviously).

I also suggest checking out QuantConnect and Quantiacs.

The tutorials are still useful to me anyway, and at least zipline is open source. I wasn't actually live trading but was considering it if I found something reliable. Wasn't able to download the quandl data into zipline but maybe someone will put up a tutorial. Would like the option to trade potentially profitable algorithms myself, otherwise it feels like the focus is performing spec work for institutional investors.

I just started to use Quantopian in the past few days and aiming to write my first algo for live deployment. I use R and external data sources and I was just thinking 'today' of the value of learning to write 'algos' here. With Robinhood, it meant there was an edge for the little guy and strategies not scalable to institutional level or strategies where commissions would erode the edge. All those were possible here, the community seemed good too. However, losing the live trading is a missing chain in the link imho. I am quite surprised that the number of people actually trading live algos is so little, very surprised actually. If it dosnt make economical sense, then heck I can see why you have to do it. It is a true danger now that you may lose many users as they may write scripts in R, python or any other external software / scripting / programming. The major win here or the major win 'was' that you could write a script and back test and utilize the same rules live. Excellent..... Very disappointed here, however, maybe something open source like this is an option? who knows but looks like crawling back into my cave. I hear zorro trader has direct broker connections, have to learn their language but might be worth a look.

Last thing I'll say here before I go on with my business, but I believe that your quote "Quantopian inspires talented people everywhere to write investment algorithms" no longer stands true as there is very little incentive for low capital and less experienced traders to actually use this website. Our algorithms serve for very little personal gain after the removal of live trading and the only gain we could possibly seek to make requires your validation, where facts show the chances of actually winning a contest are abysmal, even with an impressive algorithm. Not everyone is capable of developing groundbreaking algorithms, which you guys are indirectly forcing, but everyone is capable of learning something new.

My basic assessment here is that Quantopian gave brokerage integrations a try and it didn't work out. After 3.5 years of working it pretty hard, publicizing a lot, and even after offering totally free trading, out of 100,000 registered users, only a few hundred are impacted by this change. If retail algorithmic trading were a way to print money, then I would have expected much greater adoption. It would seem that the market just isn't there. It was a pretty expensive way to get the answer, but I guess that's for the Quantopian VC's to decide if it was worth it.

It is a minor disappointment for me, since I do not trade with Quantopian (nor anywhere else, at this point). I was hoping to make a little money, either with the contest or by getting an allocation, and then put it toward an algorithm.

It would be interesting to know how Quantopian real-money users fared over the last 3.5 years. It may be that Quantopian, on average, is doing users a favor with this change. Are there data that could be shared?

@fawce This is a terrible decision. The 99% of users who don't get an allocation will leave the platform after a few months. Who's going to start working on a great futures algo unless there is the option to trade real money on your own (if you don't get an allocation?).

There's no going back here. Once you shut down the option of live trading, users will migrate to other platforms and never return. I second Viridian Hawk's suggestion. Makes us move the prime broker and charge a reasonably small monthly fee.

My 2cts: People are motivated by different things, but greed is a very important one. Applause another one. To be honest: I'm motivated by greed, and in my mind, the chance to make money with developing an algo for the contest or fund is rather small, and the potential earnings are meagre, so my best bet was to trade my own algos. I trade depending on the day between 80 and 110k, and my performance is above market with lower risk (known risk aka exposure).

Quantopian would like you to develop high capacity (millions/billions throughput) market neutral, dollar neutral algorithms using datasets other then price/volume, and it happens that if you trade those algo's with sub 500K, you probably get eaten by the transaction costs and slippage we have on IB and RH. Hence those algo's are not interesting for me now as I dont trade that much money freely

Hence for me it's easy: I'm here for my own gain, and my own gain is developing algo's and deploying real live algo's that fit my risk profile. So now this unfortunate step makes me think: why would I stay, why would I contribute? Almost no reason to do so except for some research. I might still share algo's but probably more in the form of notebooks as proof of concept...

What now? @Lecoque shared his libraries, there is IBridgePy.com and there is Zipline-Live that probably have the least amount of work for algo's that have a known set of assets as those algo's don't need pipeline. FYI: IB has a limit of 100 assets where you can get realtime quotes for (or pay for more) so algo's that have less then 100 assets, they are fine.
Or you can go for the other platforms, but they seem to be less advanced and I don't feel like rewriting 100's of Algo's

But this is also a blessing in disguise. The broker integration did not have futures, did not have options and I would love to trade those as well.

Bottomline for me: Fawce, Dan, Delaney: thanks for the ride, I guess this was hard to decide but Quantopian is not democratising the finance industry anymore.... unfortunately.

The thing that is hardest to swallow is the time we get to find another solution: it is way too short and it shows that you guys actually do not respect nor value the community as you claim to do. PLEASE GIVE US TO END OCTOBER!

I feel a bit betrayed and sad, as I believed in your mission... I guess everybody needs a sucker punch now and then

I'll share a few more algo's for the benefit of the community but for the rest: I'm out, no more hours on the forums, no more helping people with Q worthy algo's
I rather put my energy in zipline-live or other initiatives

It would be interesting to know how Quantopian real-money users faired
over the last 3.5 years. It may be that Quantopian, on average, is
doing users a favor with this change. Are there data that could be
shared?

@Grant -- coming from a start-up background myself I can tell you that if we had success stories on our platform we did our best to publicize them. Since they haven't I think your suspicion may be correct. :)

I think the reality is -- and you see this in the contest as well -- this stuff is f---ing hard. It's surprisingly hard to have a hedged, consistent, low volatility algo that beats the S&P500. I mean we're at a disadvantage compared to Wall Street, but look even at hedge funds -- they're struggling. Often these funds are losing money right now in a more or less straightforward bull market. As soon as you remove "market risk," "sector risk," etc. you just replace it with alpha-factor risk. That one is really hard to overcome -- an accidental bias or market dynamics shift and all that hedging doesn't save you.

So I think people must give up when they can't find anything that backtests consistently well. Or if they do find something they give up as soon as it fails miserably out-of-sample. Proper statistical rigour is really really hard. Or on the flip side, somebody with healthy levels of skepticism may lack confidence to live-trade their algo. And once you factor in slippage, commissions, margin fees, and taxes, well... buy-and-hold VOO or VT might just be the better option.

I'm surprised though. I have seen some really promising looking algorithms in these forums (and I assume there are a lot of people who wouldn't dream of posting their most promising algos -- I know I haven't). But you see some of these great algos and they've been cloned hundreds of times. I would have thought there'd be an army of lurkers somewhere live-trading these. I'm surprised to hear there are only a couple hundred. Maybe 200 is the total intersection of world population, python programmers, interested in the stock market, and clever. :)

Namely, wouldn't Robinhood find this aligns well with their business
model?

@Stephen Hanly -- to the contrary, my experience was that Robinhood was adamantly against in helping me get a second account running for use with Quantopian. I mean here I am trying to deposit a hundred thou into their platform, and they're just like no. Don't know what's up with Robinhood, but I did not get a good vibe from them.

I would definetly pay for the live trading if that was an option. I have invested alo, ALOT of days work on this tool, it would be totally dissapointing to see the tool go :(

Anyway, I will probably just move to quandconnect and start recoding everything there.

Maybe 200 is the total intersection of world population, python programmers, interested in the stock market, and clever. :)

Interesting. But those are just the attributes required for successful Q hedge fund algo writers. I'd think that in order to attract prospective fund authors, it would have been a plus to offer the trading platform, but I guess it didn't work out that way. There had been so much interest in futures, then they come, and poof! Odd. I'd think it would be in the elemental nature of serious traders to want to use the platform with their own capital, and also try their hands at getting a Q fund allocation. It seemed like a winning combination. Overall, given only 200 real-money traders, it suggests that the present pool of serious Q fund algo authors is rather small, but it is not clear that dropping broker integrations will help the numbers grow.

I might get some backlash for these comments I think this is a good move.

Quantopian has positioned itself as a crowd sourced hedge fund and moved away from the broker model. Why would a hedge fund provide brokerage services?

Secondly it is not difficult to get an allocation and the allocations are not 500k but millions of dollars which means a decent share of the profits. I hope that now that they are more focused we will see better opportunities for some of us who are keen on making a career in quant finance by introducing new datasets, improving speed of backtests and more tutorials on clustering etc.

Why do I feel like this isn't adding up? The entire notion of algorithmic trading is to be able to utilize the algorithm. What drove people to learn the platform in the first place was the ability to identify strategies where they could invest their money, and have a computer, which is far more disciplined, execute a specific strategy. This is beyond stupid, and no rational business person in their right mind would make such a decision, unless there was something else behind the scenes that we're not being told. I see this as akin to the pattern day trading rule. They say it's designed to help the average consumer not get taken advantage of, but the reality is it's designed to limit average consumers ability to profit from the market, because any idiot with half a brain can beat the average annual 6% return. You guys totally sold us out! I wouldn't be surprised if the other sites were next. "Level Playing Field", yeah right. You know the day you set out to do that you would be a target, and instead of standing up for the little guy, you totally sold us out. So how about you change the "Level playing field" part to "Playing field", cause that's exactly what this is now. One big joke. The question is, did they pay you to sell out, or threaten you.

More liberals complaining about free stuff getting taken away from them.

I haven't been able to develop an algorithm that would receive an allocation, but thanks to the community ideas discussed, I have been able to develop an algorithm that I have been profitably trading since last December, while I continue to investigate new ideas. What attracted me to Quantopian in the first place was the complete integration between backtesting and live trading via IB -- Had the IB not been present, I would have looked for other options or stayed with R...
Joining Quantopian just for the hope of getting an allocation would have only attracted good quants, but then the fora would be thiner with fewer ideas and contributions -- this as others before me said may end up reducing the uptake in the quants Quantopian says it wants to attract.

I would also happily pay $20-30/month for the ability to trade a couple algorithms integrated with IB, as is today.

@Aqua Rooster: "Secondly it is not difficult to get an allocation " ... I beg to differ, the threshold is pretty high and although I wrote 100's of original algo's, none were selected, but maybe you can help in the true spirit of communty: can you share an algo that has the basis of something that could attract an allocation? There have been only a few allocations and there are 100's of live traders and 1000's of tinkerers... the % of success is pretty small

@Miles Adkins: If you have nothing to add to the discussion please go [email protected] somewhere else

@Peter, I get the feeling that they are looking for ultra low risk algorithms. Something that won't lose more than 2-3% over past 10 years.

Please keep live trading feature and consider to monthly charge .

Damn it's been real

Very disappointing decision.

Quantitative processes, analytic thinking, pipeline programming, etc are incomplete without an interface to the markets. From open outcry to HFT all strategies/algorithms must be able to meet the final dynamic, implementation.

"No operation extends with any certainty beyond the first encounter with the main body of the enemy" - Helmuth Von Moltke

Definitely a challenging decision for all those who have invested time learning the specifics of the Quantopian system with the full intention of transforming learning into live trading.

For those of us that algo trade with Robinhood, can someone provide an alternative algo trading platform we can use and provide step-by-step instructions on how to get it to work with our existing algos?

A monthly fee or even a percentage of our profits would acceptable if live trading with Robinhood was possible. I am in the process of learning right now but definitely want to put my money into any algorithm I develop that does well. The criteria for getting an allocation for your algorithm is very restrictive (although I see why Quantopian would want it that way). For the moment I would be happy with any algorithm with decent alpha. A longer term goal would be to develop an algorithm that could get an allocation. I don' think there is much in it for me anymore if I can't try to profit with my own money until I figure out how to make an algorithm that gets funded. I am realistic in that it will take me a while to become an algorithmic investing genius. I can't risk my time and effort to something that has such an uncertain return. I make decent profits in manual investing. I believe in the long run I could make an algorithm that fits Quantopian's requirements, however, why risk trusting them? This is a major change in their promise. How do you even know they don't use some people's algorithms without paying them? They have to do what is good for their business but we have to do the same. There is simply too much risk for any beginner to commit time to Quantopian now.

So if i write an algorithm under Quantopian that is not qualified for allocation and i want to use my own real money to trade on such algorithm, what choice do I have going forward ?

@Kory Hoang
Seems you have experience on many other platforms and software. Could you please open a new post and talk about it? In this way all of interesting user here can put their question there.

While I am also very disappointed in Quantopian's decision, I think we all have to accept that Quantopian is a financial enterprise. Having 300-400 accounts at USD 30 per month just doesn't make sense for any business, especially if there's no real growth relative to the effort and resources put into it (just take a look at how much a good programmer is paid in Boston).

I agree that 5.5 weeks isn't enough time for most of us to properly test migrated code (I would have thought that a minimum of 2 weeks of paper trading and 2 weeks of trading with small money would be required before safely redeploying full capital). I have played around with Zipline-live and I have installed it with little difficulty on Mac OS X. Looking at Github, main limitations currently seem to be:

  • limited to having no pipeline (a workaround is in the works but it may not be ready by October 1st)
  • limited to 100 tickers (I believe you can pay IB to increase your data priviliges)
  • multi-account support doesn't seem to be there yet (this is a problem for those with F accounts)

Anyhow, I think most us need to move past the bitterness displayed in some of these posts, and find the best way to move forward in a productive manner. Zipline-live is the clear way forward. We should be in able to get something that can run our existing code with minimal changes. This would allow us to continue using the Quantopian Backtester to test new strategies, paper trade them for a few weeks on Quantopian, and then redeploy them with minimal coding changes on local instances of zipline-live.

Of course, running a cloud-based solution was super convenient, and problems were quickly dealt with by Quantopian's team. We will now have to double check that our connectivity and algos are running properly every time the market opens. Frankly, this is not a big deal.

I run 2 real money IB accounts traded though Quantopian. My plan is to buy 2 mini PCs for USD 100 each (https://www.amazon.com/Azulle-Quantum-Access-Windows-storage/dp/B00X4O6GRK/ref=sr_1_1?ie=UTF8&qid=1503469611&sr=8-1&keywords=azulle) and install python2/virtualenv/zipline-live on them. These mini PCs take no space, consume virtually no energy and they seem relatively stable (they are slow and have very slow write speeds, but that's a trade-off I can live with). If such a solution works, it will be relatively inexpensive and hassle-free to run your own algos with zipline-live...

@Mattias Lamotte , sorry I may be asking a silly question. Does zipline-live provide real-time market data (the same as Quantopian) ? The reason why i am asking is : I try to open a new account at interactive broker and i know that we need to pay for market data fee in case i need real time market data. Without paying such fee my stock quote will be in "delay" mode. So i am just wondering how zipline-live can provide real-time market data ? Also, what is the coverage at zipline-live ? Only US stock ? How about CMEGROUP, currency, gold/oil ..etc ? Thanks.

@Tony Chang Zipline is the library on which the Quantopian backtester is based.

Like Quantopian, Zipline-live doesn't provide any real-time data. Historical daily data can be pulled from a variety of free sources (quandl, yahoo finance, google finance etc), Real-time data needs to be bought from the broker in any case (no free lunch).

While I am not aware of the existing capabilities (I believe the first iteration will be limited to US stocks), I believe that zipline-live could easily be configured to work with any financial product. I suggest you go explore the Github pages of zipline-live to get a better sense of capabilities.

@Mattias Lamotte, I agree with you that Zipline-live is a way forward, at least that would be my preferred way to go. Unfortunately it doesn't have all the features available on Quantopian (yet?) nor we have a way to make use of Quantopian data with it. Those are the main issues in my opinion. Having to host our own Zipline-live server is just annoying but not a major problem.

For example, Does zipline-live support morningstar stuff ? Or the US500 stocks ?

@Tony Chang and @Luca, it looks like all the morningstar stuff (fundamental data and all Q500US and Q1500US universes) are not going to work on zipline-live. Looking at the Github pages, they are working on a pipeline workaround where you will be able to download all the tickers for a given exchange, and then set filter parameters (e.g.: moving average of dollar volume) to filter out illiquid names from the pipeline output dataframe. I'm not sure this will be ready by October 1st, and in any event it will represent a change in code logic that will require a period of testing/debugging to adapt to zipline-live.

Interactive Brokers do have APIs to call some fundamental data (https://interactivebrokers.github.io/tws-api/reuters_fundamentals.html#gsc.tab=0) but I don't know if historical fundamental data is available.

I am relatively new to the Quantopian platform but am not sure I will continue at this point. I don't fall into those couple hundred people actually risking their own capital, but I certainly identify with them. I always saw live-trading as a stepping-stone on the path to achieving an allocation. What if I develop an algorithm that doesn't scale well or doesn't meet Quantopian's narrow risk parameters? To Quantopian, it's useless but to me, it's potentially supremely valuable. Oh well. I guess I should go build skills elsewhere and then return here when I feel I can compete on the long-short, market-neutral playing field.

Very frustrated. Just started a month ago. Hopefully I'll be able to make the transition to zipline-live, but without RH integration it's fairly useless to me as-is, so I'm hoping a LOT of angry people jump into development and speed it forward as quickly as possible. I would also have supported paying Quantopian, but I guess it isn't worth their time.

It would be really, really great would be if Quantopian could detail further what they meant by "We've talked to the project leaders, and we've agreed to support the future development of zipline-live." I'd honestly like to see a commitment from Quantopian to help zipline-live get as many of Quantopian 's features as possible operational ready by the time Quantopian shuts down brokerage integrations. If you can do that, it'd be a relatively seamless transition and zipline-live developers could add additional features after that on their own in the future based on the support that Quantopian provided in getting them up to speed.

And frankly, a ~30-day shutdown window is a bit short. Not that I expect anyone (in any industry, ever, anywhere) to admit mistakes or change decisions based on user input, but Quantopian should seriously consider extending that to give zipline-live more time to rapidly develop, especially if Quantopian weren't actually serious about supporting their development.

@ Fawce, Dan Dunn -

Presumably, you are still using a feed from NxCore, and will continue to use it for the Q fund? Do you see any path to supporting the crowd with the tools to create a zipline-live compatible OHLCV minute bar feed? Would you be willing to open-source your injestor architecture and code? Or maybe even supply the feed (with a volume discount passed on to the crowd)?

I'm just wondering if there is any path at all for your abandoned crowd to obtain their own 1:1 compatible data, so that Quantopian research/backtesting/paper trading could be used, and then the algo ported over to zipline-live?

By the way, although I'm sure there were compelling reasons for the short notice, 30 days notice is pretty short for such a major move. Without additional background, I'm compelled to preach. In the business world, you'll have trouble if you continue to apply this approach to stakeholders (e.g. Q fund "managers," VCs, Point 72, prospective customers, vendors, etc.). You'll get a bad reputation and folks won't want to do business with you. Let's say you allocate $50 million to a user algo, things are going great, and then the author decides to pull the plug, with only 30 days notice? You'd likely not want to work with that guy any more. It does feel like certain elements of the Golden Rule were overlooked here.

On a separate note, it is perplexing that the retail trading industry is huge, yet there is no market for enabling retail traders to do a bang-up job of it. I think this was the vision expressed in Fawce's Quantopian Manifesto. Build it and they will come. Make the market. Maybe the market is primarily gambling? The idea of thoroughly researching a strategy, coding, debugging, and deploying, doing out-of-sample paper trading, etc. just doesn't work at the retail level? I'd think that IB and RH would be super disappointed, but maybe they already understand that the upside potential is just not there.

I suggest a techincal solution such that users can continue trading with Q. If Q could implement a new trading option, in which case the Q servers would establish a connection with a software client residing on the users PC. That client would in turn be connected to the users IB TWS. The IB TWS has a well documented API. The software client would act as a bridge between the Q servers and IB TWS. The Q servers would send the purchase requests as calculated by the algo running on the Q server and receive back portfolio information.

I hope this can help.

Thanks

Luc

For any IB users who are disappointed by this news, please check out the platform IBridgePy. It is a platform that (judging by the name) bridges IB with a python program that you build, compile, and run on your own machine. This eliminates a lot of Quantopian's fundamental problems and gives you much tighter control of what goes on in your algorithms.

Also, it has a "run like Quantopian" mode that uses Quantopian's clock rules and whatnot. The main advantage of this is being able to write and backtest code in Quantopian's web client, then copy/paste the code into IBridgePy.

It is still fairly new and in active development, so it probably is not without its bugs. But honestly, it's the next best thing in my opinion. Only big drawback is that you need a dedicated, on 24/7, machine to run it if you want to live trade.

John, the client software I am suggesting actually would use the IBridgePy python library for its implementation. It would need a feed from the Q servers to get the orders (i.e. the result of the algo). IBridgePY is the python library that implement the IB TWS API.

So can somebody provide a comparison in between IBridgePy and zipline-live ? Is IBridgePy provided by IB ? Which tool is "more long-term" ? I do not want to spend another time cycle for learning yet another platform (assuming that i need to say goodbye to quantopian).

Maybe I am mistaken for the name of python API library to to IB TWS, maybe it is not Ibridgepy.

Luc,

I don't really see the point of maintaining a connection with Q anymore. I always assumed the recommended method for using IB as a live trading platform was to maintain two connected IB accounts, one as the "live trader" (which handles the money) and one as the "observer" (which merely exists so that you can logon and use your IB account to its full advantage). If you don't have an observer, you can't logon to IB for any reason without disconnecting your live trader.

Quantopian is (was) able to act as the observer, but it was very limited in terms of what it could look at, in my opinion. IBridgePy is a very, very good solution as long as you are willing to put in the extra work.

Tony,

I don't know much about zipline-live, but in the small snapshot I've got of it from this thread, it seems much more limited than IBridgePy.

IBridgePy will let you look at any ticker (stocks, futures, options, forex, whatever) that your IB account has permission to see, and lets you trade using whatever method you want, provided IB TWS has the capability to do so (which it should, because it is a very powerful platform).

There are almost no limitations to what you can do when it comes to IBridgePy. None that I can think of off the top of my head, anyway.

I played with IBridgePy and it seems to work. I could query my accounts and could buy a share. They implemented all main functions from Quantopian but as extra they can have a faster cycle. 1 sec instead of 1minute. Next step for me is to port an algo and take over the trading. I’ll keep you posted.

Hi John O'Leary , thank for your speedy response. I just do a quick search on IBridgePy and immediately realize at least on thing it wins Quantopian - that the Quantopian schedule_function (which is limited to US stock trading hour) is no longer an issue at IBridgePy ^_^. This is important in case we trade thing..like CMEGROUP future such as gold / oil. Now i have 2 more questions. As you say, IBridgePy allows us to trade ANY product that IB provides right...so can i trade Hong Kong stocks (or even HK index future / option) using IBridgePy ? Also, do you know whether i can get delayed market data free of charge (or i must subscribe first even for accessing delayed data) ?
Thanks.

Also, I've had very good success with using IBridgePy to grab historical price data (any timeframe, any resolution) that can then be used for backtesting on your own backtesting platform. Again, this applies to any stock, future, option, or forex. It's important to note that this data comes directly from IB, not from some database with potentially incorrect data. This was one of the biggest limitations of Q, in my opinion. The ability to export data to CSV is huge.

Tony,

You need to check your IB permissions to see what you have access to trade. If you have the permission to trade and view real-time data for the HK stock/option/future exchange, there shouldn't be anything stopping you from using IBridgePy to live trade on them.

And to your second question, I don't believe so. Trying to access data (even if it's delayed) that you don't have permission to trade has resulted in some kind of "Lacking permission to view this data" error. There may be a workaround I'm unaware of, however.

Has Quantopian considered selling the backend software that supported the live trading piece? I wonder if there is a business opportunity to build a Quantopian as a marketplace? I had asked Dann a while ago about buying the backend software that integrates with Quantopian and individual brokerage accounts. I’m still interested in this. I would think that Quantopian’s investors would be interested in getting something for this software/interface that they’ve already developed?

https://www.quantopian.com/posts/quantopian-as-a-marketplace

I'm studying right now, and I have spent a year developing on your platform. I believe it is great and provides exactly what we need for live trading. I believe the quantopian team is making the wrong decision here in shutting down live trading instead of monetizing their service. I believe many people would pay for broker integration on this platform. I will post a petition and speak with someone in quantopian to see if it is a monetary issue, so if possible we can start a kickstarter/gofundme of sorts and see if they will implement a subscription based platform for us to use. Another possible route is to add a donate option on their website.

There are many avenues they could of taken to continue the service, and it makes me wonder why they haven't utilized those options. If they shut it down for good, it makes me wonder what their intentions are. Meanwhile, I will look into other alternatives.

@ Stephen, I replied to your market place idea or some variation thereof. https://www.quantopian.com/posts/quantopian-as-a-marketplace

It's doable if the "trading middleware" between Q and the broker(s) moves to an institutional model of charging 1/2 penny or a penny per share. The number of live traders becomes irrelevant, but the volume matters. This is what anyone pays to clear through institutional firms like Goldman, Cantor Fitzgerald etc. It's a sliding scale based on the client volume and is rolled into the trade cost.

One would have to understand the volume behind ~200 traders currently using the platform, but it's definitely possible.

It would take considerable effort to get Zipline-Live to that stage, but perhaps Q would partner in such a venture with the contribution of existing code base for live trading in return for a split of revenue. It doesn't make sense to completely drop something if there is a chance to receive some benefit by licensing it out and not have to expend additional resources. I think there are some options to explore.

How common is it for hedge funds to be completely algorithm-dependent? Is it fair to describe Quantopian as a crowd-sourced hedge fund?

Surely there must be other similar business models that could succeed without the same approval process that Quantopian requires. I would be very, very interested in keeping track of some of these younger "crowd-sourced" hedge funds and seeing what comes of them.

Is it fair to describe Quantopian as a crowd-sourced hedge fund?

In my opinion, no. Yes, the licensed algos come from a global crowd, but the business is not crowd-based as a cooperative (e.g. https://en.wikipedia.org/wiki/Cooperative). The crowd has no power, no right to information, no nothing. By design, this is how Quantopian is set up and I suspect all other so-called "crowd-sourced" funds are the same. Until somebody actually sets one up that has a fundamentally different business model, I wouldn't expect a different outcome. I don't fault the Quantopian team and their backers. They are clever, and know that they could do things differently. I have to wonder, though, if they fully appreciate the engagement they could have under a true crowd-sourced (i.e. crowd-owned and crowd-controlled) collective effort. Leveling Wall Street needs to start with the fundamentals of the business that intends to do the leveling. One draw to Quantopian for me was the hope of not working for The Man (I do that already), but that isn't working out so well. Everything seems strongly vectored in the opposite direction.

Would Pay. Big disappointment.

Quantopian Goal = Investor Client Goal

Broker Integrations = Community Goal

Investor Client Goal > Community Goal

Quantopian Goal = $ NOT Community.

Grant,

Good points. From what I could gather by looking at other similar ventures, there is some serious red tape in regards to the world of hedge funds and how they take and disperse money to their investors, especially if most of their clients are small-time. Makes a more discretely-managed fund hard to create. "Leveling Wall Street" isn't going to happen unless the regulations change.

Even so, I'm confident that some new business will emerge that will make better use of crowd-sourcing algorithms for profit, with hopefully a more attractive start than Quantopian. I agree with Grant, I don't like the direction this is going.

Disclaimer: I don't doubt that Quantopian is going to be wildly successful. It's just not what I'm looking for anymore.

The decision Quantopian made is a very difficult one. But, I think, they had to look at the hard numbers and where they wanted to go. The numbers were not viable. So, they took a step back, paused. They might reconsider later, or permanently accept the change to their orientation. They too have to evolve, do the best they can for their people.

I came back to Quantopian last year for three things. To have access to minute US stock market data at low cost, find people who shared my interest: designing worthwhile automated stock trading strategies, and prove to myself that I could reach higher long term performance levels by increasing trading activity using my methdology.

We all know it takes a lot to design interesting strategies. When you throw them in the data fire, they do not all survive. And whatever strategy is designed, it has no chance of going live unless it has passed our individual acid tests. These all seem to end up with words like: profit, edge, alpha, or this no good gizmo thingy.

We have a research platform, did I say it was free, that allows us to test whatever we have in mind. We have access to extensive historical databases. We can code, within limits, billions upon billions of possible trading scenarios.

I will stick around. Continue to use what is made available. Develop strategies with high potential, put them to the test. Port the EOD scenarios I have developed elsewhere to the minute level. Let those concepts prove themselves, and then find ways to port them elsewhere.

What I want is the methodology, the trading procedures to be implemented. Even if I have to convert them to manual afterwards, at least I would have proven to myself, using the Quantopian environment, that it would have worked in the past for an extensive period of time giving me enough confidence in my work to carry it forward to its logical conclusion. Even if it is on another machine using another programming language.

So, yes, I will stick around, for the time being.

Disappointing doesn't really even begin to cover it..

Quantopian is essentially now just going to be in the business of harvesting peoples' good work for free (via "contests") and making money off of it. Nice thinking.

We have a research platform, did I say it was free, that allows us to test whatever we have in mind. We have access to extensive historical databases. We can code, within limits, billions upon billions of possible trading scenarios.

Guy,

While this is true, it does not need to be exclusive to Quantopian. The only advantage Quantopian has now is that it's easier to utilize than other methods. If you're willing to invest time and effort, platforms like Zipline-live and IBridgePy can offer the same exact functionality and more, with the added benefit of actually still being able to live trade, all for free. And if you ask me, the limitations of Quantopian's platform are too restrictive anyway.

The learning curve is high, but that's the price that you need to pay. Otherwise, you'll need to be shelling out actual money for a decent platform that can perform research, backtest, and live trade.

If anyone is interested, I'm working with some people on potentially developing an alternative.

Send me an email at [email protected] or message me here if you're interested in being part of it or contributing to the development in some fashion. Thanks!

If anyone knows of some economical feed for minutely/secondly prices, I think some of us including myself can build something interesting (BYO-VM, options trading).

This sucks

@Dan Dunn

I want Zipelinelive to work.

https://www.quantopian.com/data?type=free - This is a nice list.

example:

from quantopian.pipeline import Pipeline
from quantopian.pipeline.data.builtin import USEquityPricing
from quantopian.pipeline.filters.morningstar import Q1500US

Will this example work in zipeline? What data types from the "free" data will be ported over for the new open source project? Is there any path for users who use the paid data to incorporate it in the Zipelinelive? I think its important to scope the branch to the larger community so we can start to plan.

Zipelinelive may be a win for us individual traders in the long run. It will provide a stable platform where the rug can not be pulled out from under us. Im sorry about being upset with management yesterday. I have spent too much energy and time here to stop now. " better to have loved and lost than never loved at all" - @ Burrito Dan

@Daniel Koh

I dabbled with QuantConnect, but didn't really care for their platform, but they have the granularity you're talking about. Quantopian was much better on all counts with what it offered, with the exception of not having second and tick data. They integrate with IB as well, but not with Robinhood, so you will need a much larger sum to begin trading, though I'm not sure about some of the other brokers. Personally, I'm going to go with SureTrader and simply trade the market manually. As far as live trading goes, none of this make business sense. When you have a product that people want, you sell it. You don't destroy it, unless you have another motive behind what you're doing. I bet you wouldn't have to dig very deep to find out the truth, but I could spend that time coding with NinjaTrader or something else. Quantopian id dead, suspiciously just like Yahoo's historical API. I guess all those sick people who wanted to monopolize data that should be free, and sell it to the public for a profit, actually won over. ;) Let's all sell our souls for jelly rolls.

"simply trade the market manually" - no fking thanks man....

Quantopian Team,

Can we get clarity from Quantopian about what exactly will remain, that is publicly available/exposed, after this change? Some questions to answer (I'm certain this is not an exhaustive list):

  • Will backtesting tools and notebooks still exist?
  • Will there still be the same(as of today) data available in backtesting and notebooks?
  • Will Quantopian Live paper trading still exist?
  • Are there planned changes to remove capability from the Quantopian API (that is not related to live trading with a broker)?
  • Is all that will remain a 'community' (similar to Stackoverflow) which is merely a forum rather than a forum AND toolkit?

There is some FUD due to the announcement without the clarity of what will still be available to community members.

Does Quantiacs permit live trading links?

Hi @John,

You said "The main advantage of this is being able to write and backtest code in Quantopian's web client, then copy/paste the code into IBridgePy.". Are you sure one can also use something like the pipeline by QuantOpian there? Besides, I use quite often or quite a lot of the Built-in or CustomFactor in my algos. I fear that by IBridgePy you can't do that.

Actually it'll still work out for those of us who do live trades, if we can get a trade notification email (so that we can trade manually) instead of actually executing a trade through a broker.

@Tom,

I've heard that by Quantiacs you can not trade equuities but futures, right?

@ Daniel,

Except, it will be delayed 15 minutes.

Daniel,

Will they permit an email out? I thought they had that portion of software suppressed to allow any outside signalling.

Disappointed by this decision. Will be spending less time on the platform now that broker integration is no longer supported. I imagine others will be too....

Very very disappointed, I do have live algos running, and I pretty happy with them.
Maybe we have to switch to something else eventually, so does anyone know which one is better:
IBridgePy, zipline-live, QuantConnect ?

@Serge

I do not know what guards they have in place in terms of outside messaging, but an engineer will probably need to implement something if they choose to take this route of emailing trades.

Correct me if I am wrong. I think the only api we could use to communicate with outside is fetch_csv, and fetch_csv could only be used in initialize. Quantpian does this for safety purpose?

Oh well. I switched to Quantconnect. I can backtest 20 times on futures, equity, forex, and options before I get one backtest in QuantoPain. They allow so many brokers for a small monthly fee, and the great news is you can develop your strategies in C# and Python for free. They only charge like $10 a month for live trading.

You guys inflate your sharpe ratios on backtests to make it look like you can make a bunch of returns. There is no leverage control, and the backtest never included interest on margin. I mean the competition used to have 3x Leverage, I don't think brokers allow anything over 2x Leverage. Beta 0 doesn't make sense, why would I want to be market neutral when the market has mostly gone up over the last 80 years.

Anyways, switch to QuantConnect!!! You won't regret it!

What becomes of the datasets? Will they be included in zipline-live, or can they be accessed via IBridgePy, Quantiacs and so on?

I feel like Quantopian just committed e-suicide here, and everyone's scrambling to work out what we're going to do next.

@Lucas Lee, @Daniel Koh, @Serge d'Adesky Just write a greesemonkey script to monitor the live-trade panel and send you emails --- or better yet, code it to place the trades for you via your broker's API endpoints. Why would Quantopian add functionality like this for you? They've expressed their motivations already. They don't want to have to support features that distract from their goals.

@Viridian Hawk , thank you for your reply. It's the last thing I might do if I cannot find a better solution.
I spent many hours working on my algos, I don't want to spend extra time to port my code to quantconnect, or setup and maintain a new system for my live trading. Unfortunately, it seems that I have to accept the change.

I think everybody who is posting about all the extra things Quantopian should do to create complex hacks, stopgaps and workarounds are missing the point. (Except me when I suggested Q give us access to the prime broker they use, because that would be the most awesome solution.) They're not going to spend their time and energy developing a workaround new from scratch that would be harder than simply maintaining the current system. Why would they do that? They are moving on from this so that they can stop spending time on it, not so that they can spend more time on it.

Quantopian clearly wants a singular focus on the Q Fund here. They see everything else as a distraction. They're trying to redefine their userbase and community along those lines.

I think everybody who is being left out of that should just check out QuantConnect. Sure it's not a drop-in replacement, but I think for most of the strategies people have been running here, you can replicate the exact same logic there. Perhaps they can make a deal with Robinhood sooner rather than later and get that integration going. They are also powered by open source, so if anybody wants to contribute a Robinhood integration, that might be all that's needed to get it going.

zipline-live is less promising. Depending on your strategy and if you can get the data needed for it, maybe you can get it to work for you. If your strategy simply works based on daily resolution OHLC bars, it could be fine.

@Daniel Koh, you could write a Greasemonkey script to do that for you from the paper trade console.
@Tim Decker, I think they answered at least half of those questions already. Everything pertaining to the contest and the fund remains.
@Stephen Hanly, I don't think the live-trading part is that hard. Sure, it's a hassle to maintain if the broker keeps breaking things, but getting Q's current code base for that wouldn't help with that issue anyways. The hard part to replicate is the data that drives the algorithms here. That's an engineering nightmare and prohibitively expensive. (I also assume that those integrations also involved business deals, which the source code isn't going to help you with.) Here is the IB API: https://www.interactivebrokers.com/en/index.php?f=5041 Here is an unofficial Python API for Robinhood: https://github.com/Jamonek/Robinhood

It is a shame Quantopian didn't spin the live-trading aspect out into a different website -- Powered by the same backend, but with different interfaces and different communities. Lets face it -- live-trading on Q was just kind of tacked on. There's so much room for innovation and improvement -- but it was never their focus so it was never properly nurtured. I think that could explain why after a couple years there weren't more than a couple hundred active live-trade users.

Hey, everybody. There's a lot of passion here. I just posted a new thread with some questions, partially out of my own curiosity, but also because I thought it might be useful for everybody to be able to tell their own story, and work from there to find solutions:

https://www.quantopian.com/posts/questionnaire-for-quantopian-live-brokerage-traders

Install Kali Linux. Reverse engineer the Robin Hood platform. Install bluestacks and a sniffer and use the SSL keys to decrypt the protocol stream. This will tell you how orders are placed. Recreate the classes to provide the same functionality we got from Quantopian, minus all the hocus pocus. You could even improve the granularity. Use Quantopian for backtesting your strategy. Use your own code for placing and managing orders. Think about it. What sort of things or we looking to do?

Create Limit and Market Orders
Cancel Orders
Create Stop Orders
Create Stop Limit Orders
Monitor Price

You could code this in a month or less, and give it way for free just to piss people off. I had started down that road when I created Vicentex.com. Then all of a sudden, upon code complete, Yahoo discontinued its historical API and made it impossible to use YQL. That's how I ended up here. Fact is, these guys work really hard to keep us in the dark, because the truth is, this is a multi-billion dollar business. Think about it. You invest your money with a broker, and you're lucky if you get ten percent, but we all know that we can blow that out of the water now don't we? So who needs brokers if we all start down this path? This is about money. Huge firms, who shall remain nameless, are replacing humans with machines as we speak, but these guys don't want us getting rich off our own intellect. Oh no. That would actually "Level" the playing field. Unfortunately for them, they can't stop us. And no, we're not going to build algorithms for millionaires and hope they fund us. Let them build their own. Better yet, tell them go find a good broker. If we can't profit from our own hard, work, why should we contribute to income inequality without an opportunity to capitalize? Sorry, gotta go blue pill on this one. Your matrix has nothing offer.

L. Williams, there is already a robinhood API found out by users that uses python to place orders. The issue as you said is data. I wonder if Quantopian would be willing to supply us data for zipline live as a measure of good faith.

@ Eric Bell, 3 months ago when i started to learn algo trade, i made a comparsion on both quantopian and quantconnect. Eventually i picked quantopian because i have zero knowledge on C# "class" (and so C# is too difficult for me to learn). Correct me if i am wrong. I believe python is more "conventional" and use the normal programming logic. C# is something hard for beginners, especially everything is in "class-style" and the programming logic flow is not "sequential" (although i agree that it is even more powerful if we fully understand C#). However, i believe most algo traders are not "IT guys". We want to pay focus on algo rather than learning programming language. That is, if both python and C# can do the same thing, i prefere to learn python because it is much easier to learn.

So do you feel learning C# is difficult ? any guidance on such learning ? i am now struggling in between http://www.ibridgepy.com/ and quantconnect.

That would certainly be an offering worth of repentance, but what would be in it for them? That way to a man's heart is not through his conscience Luke. It's through his desire. They would never acquiesce to such noble endeavors. Of course, you could use their platform to feed yours, but alas, I've said too much already. The force is not with us on this one.

@Luke Izla, Could you share any information/source of the Robinhood API that uses python to place orders which you mention above?
Thanks

Here's a proposal community. The API that exists allows you to pull data in "from" a spreadsheet via a URL. If the wonderful folks at Quantopian could spend a small amount of time extending this functionality to allow us to export our pipeline results by posting TO a URL, we could leverage the existing platform by importing the output into our own, and utilizing the data wherever we like. We could run this daily, to extract our securities, import them into our newly created portfolio management platform, and execute trades, however we like. This way Quantopian doesn't have to worry about bugs in their code, and we can leverage much of the existing algorithms, that we've painfully created for "leveling the playing field." What say ye gentlemen? This would help you keep your stellar reputation in tact, and your sins would all be forgiven.

That is never going to happen. Quantopian provides us with neatly packaged data all for free because we have to use their interface. They're not about to let users export their very expensive data for free.

Are you serious? Why would I continue to use this platform with no live trading support? huge disappointment. moving on to the next best thing.

@Tony Chang -- My experience with Python is limited to Quantopian, but I will agree, as it is typically used here it's linear, or what is called "Functional Programming." While this is simpler to learn, the other common technique "Object Oriented Programming" is considered the best practice. It's harder to wrap your head around, but in the long run it's cleaner and more maintainable. Fear not though, QuantConnect also supports Python, and their platform is pretty similar to Quantopian, and you can code in whatever style you prefer.
@L. Williams -- nobody need go through all that trouble reverse engineering the Robinhood API. It's well documented if you search Google for the Robinhood API, it's on GitHub. On your other point, just like you can't buy an mp3 or a movie and then offer it for free streaming on your website, Quantopian can't let us stream or download for our own private use the data they're paying for. Also, why would they? If all you need is daily data, you can get it free from Quandl and load it into zipline.
@Chris Venne -- I already posted the link above.

Thanks again, all, for the replies. The follow-up points I made yesterday are relevant to some of the replies that came in recently: the importance of focus for Quantopian, the fact we didn't find a viable business in the broker integration, and, most importantly, our wish that we could support this and our thanks for the community's contributions. I won't repeat those points here at further length.

Several of the comments today talked about how the interest in trading personal capital might translate into algorithms that get an allocation. Like many of you who posted here, we thought that people who were intrigued by trading their own capital would go on to write algorithms that get an allocation. Unfortunately, the data doesn't confirm that. When we tried to figure out why, our conclusion was that the problems are too different. Personal traders are often trying to capture market effects, like beta-to-spy, that aren't highly valued by institutional investors. Personal investors are also optimizing for their account size. When trading personal capital, people generally write long-only or long-biased algos, and they choose strategies and assets that don't scale well. None of these things are "bad" - they're quite reasonable. But they are an aspect of our decision to end brokerage integrations.

I also wanted to confirm that sharing the live data off the Quantopian platform isn't something we can do. We license the data from other parties, and that license is limited to the Quantopian platform. Of course, much of the data we use is available for free, like Quandl and daily price/volume bar data.

A note on moderation: I've had to make a few moderation decisions on posts that violate our terms of use. Please remember that disagreement is fine, but civil tone is important. I'm also moving this post off the "featured" tab - at 160+ replies, it's become unwieldy.

Zipline-live is in alpha stage!
How can Quantopian recommend it as an alternative? Unless Quantopian cooperates with them and assists on the development, which I think is highly unlikely.
May be I will have to get back to using IB API directly as I used to do 3 years ago, given that now IB has a native Python implementation of the API. It is also worth considering deploying algos on amazon servers manually. By the way, QuantConnect is $20/mo (not $10 for the Prime plan, at least this is what I see on the screen).
I still will use Quantopian for backtesting, at least for some time.

Dan's reply actually makes sense, as always. I think the confusion stems from the motives. As a trader/investor, speaking for myself, my motive was to create algorithms not necessarily for investors, but for myself. I'm just being honest. My goal was to turn a small amount of capital into a larger sum by applying technical and/or fundamental trading principles to the filtration and execution of trades. By removing that capacity, it firmly supports Quantopian's goal, but at the sacrifice of my objectives and others like me, which may actually represent the majority of the user base. To be frank, I don't know anyone who isn't, at some level, doing this for their own benefit, even if that benefit is mutual to Quantopian. And honestly, from a business standpoint, I understand the dilemma, but you can't deny that this is a botched opportunity. I would pose this question. If you quantify the profit generated from the investor base you seek to entertain, and compare that with charging us for the feature of being able to live trade, which has greater scale? You can instantly monetize the platform and charge a monthly fee for what's currently present. Compare that with waiting for someone to develop the sort of algorithms, you yourself stated, simply weren't the bulk of what you were receiving. I think you guys should seriously consider what's on the table, and firmly weigh the risks. As they say in trading, pigs get fat. Hogs get slaughtered. I can't see any sense in leaving millions on the table, when there is virtually no damage to the goals you're pursuing. And worse case scenario, you can always decide that the compromise wasn't lucrative enough to your liking. However, to simply dismiss it, not provide it as an option, and risk a large portion of your user base leaving the Quantopian kingdom desolate, in hopes of new comers jumping on the bandwagon to code more scalable algorithms, sounds questionable, at best. At the end of the day, if this is a business, it's objective should be to make money. Let's not kid ourselves. There is no noble cause here. Worse, your competitors will entertain both. That fact alone will kill the platform.

Good point L, but in my personal experience, Dan is spot on. The goals and methods of people with the skills and means to trade their own accounts are different from those of the fund. The two sets are very likely completely disjoint.

Writing algos and firing off them into the void in hopes of an allocation is very much a "lottery ticket", and I personally judged it not worth my effort months ago. I was far more interested in a more certain and reproducible benefit from trading my own account as I saw fit.

It is likely only students and newcomers for whom the lottery effect makes sense, and so by cutting off those of us which were using the system to trade for ourselves, I very strongly suspect they've lost nothing in first order effects.

The big question is, if everyone who trades for themselves leaves, what happens to the free flow of information on the boards? Personally, I think that ended with the contests anyway, and the boards haven't been much use since, so again no great loss there either.

What you are left with, I suspect, is a feeder pipeline of students and dabblers working quietly in isolation developing lottery algos for quantopian, which would seem to be just fine for their business.

I agree with Simon. This decision probably makes much sense for Quantopian's business. Since I joined this platform I have been asking myself how could Quantopian spend so much resources on tasks that are not their direct goal, the hedge fund. I understood the need of attracting people with platform features that were meant to interest and be useful to the users and not the Q hedge fund directly, but it seems that they don't have the resources for those anymore. Ending the live trading feature means Quantopian is switching to a more selected user base whose intent is learning, researching and trying to build algorithms for the hedge fund. This might be a good move for Q, time will tell.

I don't feel betrayed by Quantopian, I didn't base my choice of using Quantopian hoping in their good intentions, even though their "open to the community" mindset helped a lot. But Q is a business and I am willing to be part of this as long as my goals are aligned with Quantopian's goals. There is still a lot to learn here and the tools they develop are open source. Also, it's a good fun trying to write algorithms that move lots of money. Something that I will never (who knows? :) do with my own money.

Shutting down the live trading is mostly a hassle for me, because I will have to spend more time dealing with live trading problem instead of focusing on developing new ideas. if zipline-live was a ready and full replacement for Quantopian platform that would be great, but it is not ready and Dun's confirmation that Quantopian cannot share their data externally makes zipline-live pointless for me. QuantConnect is then the natural choice for my live trading, but it will be a pain sharing code between two different platforms. Anyway QuantConnect is very mature and, from the live trading point of view only, it is even more advanced than Quantopian. This is because users live trading is their business, so they keep improving it.

I'm in the same boat as Simon when I started, but I did slowly move to make more Qhedgefund type of algo's where I could switch things on or off (hedging in bull markets is bull imo, dollar neutral is not possibel in Australia with IB, etc)

There are 3 things That annoys/stings me:

1) What stings me is that their mission was to democatise the finance world (this part of it) and they can do 2 things: open up their Hedgefund to the community so we can invest (I would hapopy to drop 100K), or allow people to live trade with their own money. How am I going to profit from writing high throughput algo's suiteable for their fund when the chance to profit from it personally is a lotteryticket?
2) The second thing that I think is just deplorable is the time we get to port. This REAL money, not VC play money, not LP money from Point72: I trade with post tax hard earned dough: you have to respect that and that @JohnFawcett should have fought for, either longer time, help develop an alternative or something else.
3)The third thing is that I actually paid Quantopian for the data use for research/trading, that is now completely useless as I cannot access those datasources myself outside of Q. Thats feels like daylight robery

Finally, when you live trade with you own hard earned dough, you discover that the proof is in the pudding: the executing of the algo is at least as important as the alpha you discover and I had a steep learning curve. I suspect the algo's published by non-traders, students, punters, etc will be less likely to hold up in real trading and therefor this community will be less relevant for me.

I'll hang around to find a solution but after that I will be a real leach: using the datasets for research and like that.

Interesting that there isn't much overlap of folks wanting to trade their own money, and those who show potential for writing Q-fund-worthy algos. It is "systematic" market speculation versus sober, risk-managed large portfolio construction, along the lines of A Professional Quant Equity Workflow. The retail traders with dopamine addictions probably aren't the ones interested in doing the latter work with little or no reward (the contest ranking is really the only virtual "warm fuzzy" feedback, but it is not the same as the psychological risk/reward of trading one's own money).

Part of the problem, in my opinion, is that as Simon points out, it feels like a winner-take-all lottery ("Imagine...you could be allocated $50M!"). There's no way to make individual, incremental contributions and get paid in proportion (unless one is actually employed by Quantopian). And there's no way to invest in Quantopian nor the fund. And the latest twist is that one is supposed to use data that is not available up to the present for development! And even though it was reported that it is very common (and presumably profitable) for hedge funds to use mean-reversion algos, apparently, there's limited interest on the part of Quantopian for such algos (at least that's the message I think I got). I'm not sure if "lottery" is the right metaphor, since based on my recent experience, it's impossible even to get feedback to find out if one has lost (i.e. the algo has no hope for an allocation). At least with a lottery ticket, one gets "Sorry, you are not a winner this time." I'd agree that it is way too much of a void; there's a lot of room for improvement in this area.

I'm curious about the reported recent challenging environment for hedge funds. Might this be impacting Quantopian? Certainly if volatility is low, some strategies will suffer. But then, if the market is steadily going up, with low volatility, wouldn't that be the time for institutional investors to load up on hedge funds (since volatility will inevitably come back)? Wouldn't they be re-balancing to maintain their allocations in a steady bull market, so hedge funds would see inflows (even though hedge funds aren't posting high returns)? Or do institutional investors chase returns, just like retail folks? I guess if Q is seeing flat real-money returns, it is hard to make the case that when volatility comes back, the fund returns will turn up. They need the real-money out-of-sample data, when the market turns down, and hedge funds do their job for institutional investors in smoothing out market roughness.

Disappointing doesn't really even begin to cover it. This sentiment was voiced earlier and I second it.

I've been an active member of the Q community forums for awhile now, but now saddened to say I'll be leaving because of this decision.

Q now boils down to a contest and a long shot allocation. It 's now based solely on competition. Communities are not based upon competition but rather on collaboration. There is zero incentive for me help other members because it's now a zero-sum game. If someone else wins that means I loose. Having the potential for live trading allowed me to benefit independent of any contests or allocations. Collaboration, communication, and help was a win-win. I could always benefit even if helping someone win a contest. No more.

So long (and thanks for all the fish).

Dan Whitnable

My comments on difficulties hedge funds are facing is based on articles like this that keep making the rounds as of late : http://www.businessinsider.com/jason-karps-tourbillon-hedge-fund-q2-letter-on-quants-passive-2017-8

Dan has a good point--and it's only supported by a lot of the comments here --about the disjoint between allocation-worthy algos and live traders. Looks like indeed lots of people are one or the other.

Question now moving forward is, how does Quantopian avoid becoming the Uber of hedge funds?

Uber is of course notorious for creating a technology platform that revolutionized the ride hailing industry by essentially crafting a novel way to exploit drivers. I'm hoping that with this new regrouping Quantopian can focus on better ways to incentivize and nurture their coders. (Something more than a lottery and free job training.)

Wondering if they can at least remove the 15 minute delay in paper trading....

@Dan Whitnable - thank you for your incredible contribution to our community. All the disappointment in this thread has been hard to read, but your message hit me the hardest.

I take responsibility for disrupting our community. As many in the thread have said, endowments, pensions, and other institutions face a very different investment problem than individuals.

There was a long and intense debate inside Quantopian, and ultimately I felt I had to mandate our focus on the institutional investment problem. As a kid, I won scholarships to go to high school and college. I'm deeply grateful there were endowments with investment managers planning for my future. That's why I feel that focusing on the institutional investment problem is such important work. I believe it is my purpose, and I want it to be Quantopian's. I also think it is one of the biggest business opportunities in history.

My vision for Quantopian's technology is to create THE platform for the institutional investment industry. A key part of becoming a platform is cultivating experts like you. That means we need to teach the Quantopian community about the problems institutions face, so you can help solve them. Everything in our product -- tutorials, references, examples, our APIs, the data we integrate, and the incentives we provide -- needs to support and reinforce solutions to institutions' problems.

So what's in it for you?

The entire asset management industry is experiencing an existential crisis. Over the next 5 years, I expect every investment business to undergo a technology dislocation. Some already have. Quantopian's mission is to catalyze that transition with our platform and our user community.

If we execute properly, Quantopian will be THE platform for asset management. I feel confident saying the economic benefit to you, as a community member, will be far higher with our focus on institutional investing. Your skills will be sought after, even if you never receive an allocation or win the contest. In my view, this is much more valuable to you than taking risk with your own capital.

We've been working on many amazing new platform features to go further in this direction. We've also been examining the incentives we provide. Clear purpose will allow us to focus on the open problems we have as an institutional platform -- we need to reward more community members and we need to provide more functionality.

Institutional investors need you. I hope you'll help.

thanks,
fawce

@Fawce with all due respect, your reasoning is insanely speculative and portraying this as far too noble a cause. Dan' s skill set would probably reward him with higher returns elsewhere than sticking around here until YOU guys make it so that he can get some recognition.

@Fawce,

I've asked this question before (https://www.quantopian.com/posts/investing-in-quantopians-hedge-fund) but why not look to open up investment opportunities to retail investors for the Q fund? Wouldn't that at least provide a means for us to feel better about collaborating and helping one another, if we know it's in our best interest too because we've invested?

I hate the idea of only looking to help the institutional investors, I know they move and control the most money - but the retail investor needs help too. And I fundamentally believe they are more deserving of our's or anyone's help. If there were a means for retail investors (and community members) to have access to the output of our work and contributions to the Quantopian community, I'd think we'd all feel better about helping.

@stephen, you voice my feeling well. Let us enjoy the collective work by being able to invest in the fund. I understand that one needs to be a accredited investor....There is really no incentive left for me otherwise.

@ Fawce disrupting the investment industry starts at forcing the players to look at alternatives on how to do things differently. Providing retail investors a way to do it differently would disrupt this industry even faster then etfs and Roboinvestors. Give the community more tools like AI toolkits and access to a prime brokerage would be a way imho.

@quantopian: why do we still have to pay for the datasets? Doesn’t make any sense if one can’t live trade them. Paying a lot of money for gaining a lottery ticket (change of allocation) doesn’t make any sense. I spend around 500$ on datasets which will be useless for me.

I have just tried IBridgePy and it seems to really work (with only slightly modified quantopian code) as long as you use data from either IB, yahoo finance, google finance. I would suggest exploring capabilities of pandasdatareader (https://pandas-datareader.readthedocs.io/en/latest/) for longer term daily data, as IB seems to have an issue with daily data for more than a year using the data.history() method.

This may be a viable option to zipline-live for most algos that are market data based and which trade no more than 100 stocks. The downside with both zipline-live and IBridgePy is that IB doesn't supply delayed data through their API. You either subscribe to the live data feeds (you probably need both AMEX and NASDAQ subscriptions), or you get nothing.

I think their's a good chance that both of my algos will be working in 2 weeks time on iBridgePy. Several days of testing will be needed, but I'm more optimistic of being able to make the October 1st deadline.

"First there is opportunity, then there is betrayal..." Sorry, couldn't help myself. After all the sadness expressed here.

Re: @Stephen Hanky +1 i like that idea as well. What about regulatory issues? Would investors be required to be "accredited investors"? That would leave out everybody but the rich.

So thinking about what quantopian has to offer their user base now -- let's consider an algo hedged against all the risk factors and displaying no volatility, assuming out-of-sample degradation and alpha-decay in real life performance you're not likely to beat SPY' s 8% average return. Let's say you achieve 5% returns on a $5m allocation. That's a theoretical $25,000 cut. Not bad if you can with confidence crank winners like that out on the side. But it's unclear what percentage of algorithms will get allocations. Depends also on market dynamics not shifting -- so quite a bit of luck. Hard to tell what the odds even are. But you have to divide that $25000 by whatever you guess your odds are and from there compare it to how much time you think you'll need to put into research and development on the q platform and ask yourself, is this even minimum wage?

I echo the sentiment regarding the retail investor needing more help. Quantopian was an easy bridge for the retail investor to begin trading algorithmically.
Learn from Quantopians API/ tutorials, community and get to competing. Often there are bigger hurdles for the retail investor due to high quality data being expensive, depending on back ground, difficulty with programming. Even those two points alone, Quantopian helped dramatically with the process. It kind of gave power to the little guy.

I would suggest that the "few hundred community members" who have live trading algos that are making money (myself included) form their own community with the aim of allowing other retail investors to trade off theirs....Best, Pilko.

Ditto to Pilko. Absolutely disgusted by this flagrant betrayal of the very people that made Quantopian valuable in the first place, the users. The selfish leadership of this company should be ashamed of itself, and I hope the platform withers from the exodus of betrayed users in the months to come. Feeling like a sucker for believing in this parade of 'community'.

Some might not like it, but John Fawcett is right nonetheless. We have to think big.

The greatest help Quantopian can do in democratizing investing and wealth generation is by helping fund managers grand scale. The big picture over the long term warrants this approach. It is the consequence of a simple formula: F(t) = F(0)∙(1 + r_m + α)^t.

Here is my Excel version:

A trading strategy must be scalable. It must be executable in real life. It must be sustainable, and, it must be maintainable. A short term vision of the problem is not enough. I have been saying this for years now. These properties can also help the smaller trader/investor.

Nonetheless, the problem is not so much at the individual small investor level. Every one of us can solve that problem to our liking on our own.

However, going globally, everyone could benefit from such an undertaking. Evidently, Quantopian, the funds, but mostly all the pensioners to be. Not to count endowment, heritage, nation, mutual, hedge, investment or philanthropic funds. The task is huge, you have to start somewhere. This is really like helping humanity going forward, giving them the tools to care for the aging population as well as assure younger generations of a brighter future.

If Quantopian can bring 10 alpha points or more to the plate to ensembles of large funds, they will make a difference. All these funds represent people's hopes of a decent retirement, or for others the tools to help people to better lives.

It is what will make a difference.

It is why I am sticking around. It is a cause that is bigger than I am, and I would be proud to participate if I can.

Hi Fawce -

I guess I'm just not getting it. You seem to have a vision that Quantopian will somehow be different, but its the people and their collective power that will make a difference. This new-fangled Internet thing gives you access to the global community, and it is a golden opportunity to do something fundamentally innovative as a business, to be refreshingly transparent, and solicit user input and approval at every turn. Have public-facing change and bug logs, establish a process for request for comments, publish minutes of your meetings with VCs and existing and prospective investors, publish the fund performance, and the list goes on. And when folks say, "Hey, I'd like to invest in the fund" actually task a regulatory expert to provide some options for discussion. You need to think how you'd like to be treated, what voice and real influence you would want to have if you were on the other side. I don't think you want to be the Uber of finance, as someone commented above. So think about what you could do differently so you don't lose any more Dan Whitnables. It's really not that hard. In the present case, as soon as the thought came into your head, you could have done a post, to give your precious users as early a heads-up as possible.

By the way, how much of your own money were your trading on Quantopian? Where will you move it? Or do you have access to the prime broker?

@Dan Whitnable, I always liked to read your replies to posts in the forums because I recognized that you had deep insights and understanding of the Q platform. I could always gain something new by reading your posts. I hope you will reconsider your decision because a community needs members like you who are inherently helpful in nature. If your decision is already made which I am afraid it is from the wording of your post, I would like to thank you for taking the time to answer a couple of my questions regarding pipelines (early on when I was clueless with the purpose/ intent of the pipeline) and I wish you the very best in your future endeavors.

To be honest, I came back to this thread every couple of hours just to check if we could affect the final decision. I think I am not the only one who will leave forever if Quantopian closes live trading.
Quantopian is the best site I have found for individual quantitative traders, I met so many talented guys here, I cloned their algorithms, I learned a lot and I tried my own, I spent ton of hours just to make my algos profitable, I do live trading since I want to get returns from my efforts. Now obviously, I am not the user Quantopian cares about, but how could Quantopian grow if all users like me are locked out?
I understand a company is not a charity, there's no free lunch, but there will be no lunch anymore, free or paid!

damn, RIP

@Lucas Lee thats also the reason I keep checking this thread frequently. I think that the change is final though. Extremely disappointed with the way things were handled.

So for those who focus more on individual brokage real-money trading (instead of Q contest), i spent a few days and do some crack and summarize the following possible directions :
1. Go to IB and learn their native API (i only know a little bit on python and it is really hard for me)
2. Go to Dr. Hui Liu http://www.ibridgepy.com/ and (hopefully) it works like Q and minimize our learning curve / programming overhead
3. Go to http://www.zipline-live.io/
I believe i will go to (1), although the starting learning cost is VERY high. The IB API is a bit "low-level" for me, and it can throw a lot of terrible traceback error messages on day 1. However, i believe this is a long-term effort. I do not want to rely any more third-party API (such as 2 or 3), because i am unsure whether their API development is susintable or not.
Please correct me if i am wrong (and guide me to the right direction).
Also, i will email the above to Dr Hui Liu and see how he comments as well.

The irony here is so thick.

How can a company based on live trading large sums of money not raise enough capital through its own algorithms to pay for the maintenance of an API that is effectively free online (robinhood) - But you still allow me to live trade on 15 minute delayed data as many algos as i want... The only thing that is adding up here is the that the lowest barrier to entry free trading platform just sold out to the big money.

The notion that trading for a hedge fund is better.... Seems evil. The only good side is if you can code that well you probably dont need to look too hard for a job... regardless i feel robed of the last year of my efforts... last 4 month of full time live trading and learning (backtesting is nothing like real trading)

Like this is comically bad. The Yahoo real time data going away...

I hope management knows that history will pass judgement on them...

little sun light exposes the truth...

What i would like to know if quantopian is successful in the hedge game would the go back to provide free trading again? seems like its just some server space and using old code at this point for the robinhooders at least... Seems like the meritocracy is being killed purposefully.

Unorthodox idea (on behalf of zipline-live):
why not channel this energy to make brokerage integration happen?

180ish angry comments here, almost 500 unique visitors at zipline-live.io in the last three days.
Less than 10 people got involved on commenting issues and only ONE committed to make a change.

If you want to make a difference grab a ticket and start making it happen:
https://github.com/zipline-live/zipline/issues?utf8=%E2%9C%93&q=is%3Aissue%20is%3Aopen%20label%3Anewbie

@Tibor Kiss, the problem with zipline-live (and other software packages) is the data. Even if we get a full Quantopian replacement we are left without Quantopian's data

@Tibor Kiss, no offence here, I think zipline-live could surely make live trading happen, and I'd like to contribute my efforts if needed. But there is no way for us to get reliable data source as Quantopian provides, which I think is the one of the key factor for a successful trading system.

Sorry, I didn't notice Luca has already posted the same problem I concern.

I agree with Tibor, we must put our efforts into zipline-live. Of course Luca and Lucas are also completely right! Data are essential!

For EOD price data there is quandl, while for fundamentals gurufocus.com provides an API (the data are from M* like in Q) for a reasonable fee.

We could start compling a list of the best datasource available online free or for an affordable fee also for retail investors.

Last but not least: We should not forget to be grateful to Quantopian for donating the source code and make it open! Don't get wrong: I don't like the decision to phase out the brokerage integration, too... but if the Q team has now time to add new features, as historical fundamentals and better performance in pipeline, maybe it wasn't so a bad thing!

If Q will still be open to retail traders for research, could a separate forum be set up for topics that are now deemed irrelevant and a distraction? The forum is valuable but weeding through posts on stuff that is clearly not germane to the institutional market will be unproductive. And the retail moochers would probably prefer not to sift through posts on long-short algos that will only trade > $1 M (and no retail trader would invest in regardless).

QuantConnect replied; emphasising continued support for individuals. Reality is given live & historical data costs (for 000's of assets) they're the only real alternative. https://www.quantconnect.com/blog/democratizing-finance-empowering-individuals

@fawce,

I'm new to this community, but I spent a good chunk of my career in the Linux Kernel community as a major contributor. I have seen many open source communities come and go...the one thing I do understand is how to cultivate it. Do not take the community you've built for granted. It is not easy.

If your goal is to cultivate a community to being able to address the institutional investors problems, you have to think of how you are going to get members from point A to point C. How I started (and I suspect many people here) was wanting to learn how to algo trade with their personal goal of being able to do algo trading for themselves (call that point A). We learn enough to put our first algo live and put our money on the line (point B). When money is on the line, there's a lot more skin in the game to learn and improve.

I've only done 1 algo I was going to personally trade on, but through that process I have developed the lens of how to do the next one that could be adopted to an institution (point C). And an institution algo was going to be my next focus.

Without point B, you can't get enough members to come in the first place to build the pipeline of knowledge to get to the institutional knowledge level.

IMO, it's idealistic to think people are going to come here in any meaningful way to learn about institutional algos to get a job some day. Anyone who is good wants to trade for themselves.

@luca & @lucas: I'd argue that data is not a problem. You can already use EOD data from Google / Yahoo.
If you need fundamentals then morningstar is a way to go. Maybe you're not aware, but yearly subscription for m* costs less than quantconnect.

If you need finer granularity data (for backtests) you can sign up for IQFeed, gurufocus, EODData for a very modest price.

I'm extremely disappointed in this news, but the decision is understandable. I had a feeling that there was going to be an issue down the road when the real time pricing was removed from live trading. I'm sure this stuff costs money, and it isn't worth it to Q for a few hundred people, especially those on Robinhood, who by their very nature are running strategies that don't meet the requirements that Q is looking for.

I must give a shout-out to Jared and Alex from Quntconnect who have been super helpful on Slack chat and the QC forums, I recommend all of you to try it out as a replacement.

Designing a trading strategy will most probably be a variation on a theme of something that is already out there. In the sense that it will most probably perform at about the same level.

In a way saying: F(t) = F(0) + Σ(H.*ΔP) → F(0)∙(1 + r_m)^t. That is, your trading strategy H, performance wise, will tend to about market average over the long term. It is the most expected outcome. Look at all the professionals, what you see is that most don't even outperform the averages. Yet, everyone persists in doing the same things they do, hoping to have better results.

We simply have to do better. If we want more, we will have to do more.

Just look at Vanguard if you need some kind of confirmation. It now exceeds: F(0) = $3T under management with no foreseeable alpha generation. That is a lot of people accepting the no alpha to be had scenario.

What we should want is: F(t) = F(0) + Σ(H.*ΔP) ≥ F(0)∙(1 + r_m + α)^t which is a much harder problem to solve. Especially, if we want our CAGR and our alpha to last.

We are the ones able to generate this alpha. It comes from our skills at finding new, or perfecting existing, trading procedures. It is our ability to mine and extract those alpha points that will prevail.

But, to be worthwhile, they need to persist for the duration, over the long haul. Winning a six-month contest is insignificant if the alpha can not be maintained. It is why we do all those tests in the first place.

We want to find it, make sure it is really there, and then get all the probabilistic assurances that our set of trading procedures will prevail going forward. Because that is where we will have to play. Where we need to select an unpredictable path (our trading strategy) among gazillions of other possibilities. And, we simply want to win the game for ourselves and/or for others, no matter what.

I find operating a small portfolio a total waste of time, talent, and resources. Sure, you will make some money. From all the skills I have seen in this forum, I have no doubt about that. But the question is: is it really worth it when we could consider another dimension that could produce a lot more.

Here is the same chart as in my last post with smaller stakes while maintaining its 30-year outlook:
Smaller Stakes

In terms of global wealth creation, compared to the previous chart, I would have to classify it as peanuts, at every level. All these could do is help a few individuals that can generate the alpha and have the highest initial capital. Nonetheless, there is no real grand scale benefit, no societal impact to have 200 or 1,000 more almost rich people.

Whereas, your talents could be funneled to help the majority and bring you even more than what you could have done by yourselves. For me, playing small is like seeing so much talent going to waste.

Whatever trading strategy we might design showing some sustainable alpha deserves a lot more than being limited by lack of capital. Please, find ways to remedy that. Find added capital, push the envelop.

If your alpha generating trading strategy is not suited for any Quantopian contest, it does not really matter. We can have our trading strategies work for any organization(s). Or, we could use our other talents to work, and build our own.

My advice is: don't sell yourself short. You are all worth a lot more than that. I would even yell: Quantopian Traders Unite. We have a considerable advantage over most. We can program these things. We can understand what the others are doing. Even, might find some justification for using some of those code snippets in our own programs. You know who are the best in this forum, trust, or find compatible with your own views of the markets.

Then, support them or let them support you. What you should be negotiating is this long term alpha. If you have it, others should pay you for its use. It is your intellectual property. If others have it better than you, be ready to support them and get the higher alpha. In the end, it is what will really matter: how many alpha points could you get? It is also what Quantopian is seeking.

Already, Vanguard can give you market average (r_m) at low cost. We need better than that, we deserve better than that.

The important part might not be the language, the platform or the trading environment. It is the strategy itself, and that is you and your trading skills.

Anyone can show performance snapshots of their trading strategies, provide Alphalens or Pyfilo analysis without revealing their underlying code. Support the best developers you can find in these forums. Find some common grounds, grow together, grow faster. Promote your strategy. Do more. And note that any successful trading strategy can be ported to any other programming language. All that you program is about big money, not as in the chart above, but as in the chart in the previous post.

@guy -- as I understand it clever trading doesn't create wealth, it simply redistributes it. Do we know who we are benefiting when we are awarded an allocation? Is that information public?

@Veridian, such an old cliché. You give the argumentation to stop trading altogether by defining the game as if a zero-sum game. In such a case, why even bother designing automated trading strategies? Why even use Quantopian?

The trading after-market is in fact simply redistributed wealth. You do exchange 4 quarters for a dollar. But, that is not a concern. As a trader, you just want this “redistribution” to go your way, and you would like to make sure it does.

The expression for market appreciation is: F(t) = F(0)∙(1 + r_m)^t which over the past 240 years has been up. And this is the differentiator. You exchange your money for an asset that can appreciate over time. There might not be wealth creation at the time of the initial exchange, but there can be wealth accumulation after, which is all the motivation needed to play the game.

Accepting the zero-sum game argument is equivalent to saying that there is no alpha available which is the same old academic view of the no free lunch, the portfolio efficient frontier, or the efficient market hypothesis; expressing that the most expected outcome over the long term is the average secular market return (r_m). Notice that an investor does not speak zero-sum game, he speaks wealth appreciation.

Again, accepting the zero-sum game argument has for only conclusion: stop trading, it is not worth it. It only leads to: F(t) = F(0) + Σ(H.*ΔP) → F(0)∙(1 + r_m)^t.

But this, is still wealth appreciation which you can have almost free by buying low-cost index funds. No need to trade at all. Practically no work, one decision, and then you wait for decades. The no hassle solution. Vanguard likes that very much having $3T under management. Have you ever estimated the value of one added alpha point to Vanguard's return? Well, here it is: $30 billion dollars just for year one. How much would they pay to get it?

For our share of what is out there, we should be ready to render a societal service just like the insurer selling insurance. Traders are the risk takers when others are not willing or ready to take it. And for this, they should be compensated. Each time we buy or sell some shares, the other side is telling us, right in our faces, that we are wrong to do so. It is only our expertise, or our programs, which will make us prosper or not.

Are you not more clever than the other guy at this game? Can you accumulate wealth over decades? That is the goal. You want to gamble it out, that is fine too. Someone will benefit from that, it might even be you.

I prefer trading systems. Organized and preset to do my bidding. Will accept certain types of scenarios where I don't know if I will win, but nonetheless accepting the inherent risk that goes with the trade. I will backtest over extensive periods of time to make sure it was at least possible to win over past data. I want to put all the numbers I can on my side. I want to make sure that my strategy will respect the following: F(t) = F(0) + Σ(H.*ΔP) ≥ F(0)∙(1 + r_m + α)^t. And, I want the alpha not only to be positive, but to also be higher than just average.

The program strategy designer can do more than just play for himself. He/she can design better performing strategies that can help others increase their wealth as well. And this can be done grand scale, as an added tool to shape future generations.

Give Vanguard its added one alpha point or more and see how many people it will help. Note that the alpha is also compounding.

@Guy, I'm not suggesting investing is a zero-sum game. I'm suggesting that beating by the averages it means somebody else must underperform the averages. And I don't just mean tautologically, but cause-and-effect. As per your example, give Vanguard one point alpha due to some trading sophistication, it will cause an equal drop in alpha for market participants that lack that sophistication. While you help one group of people, you hurt another group of people.

So my question was simply -- whose money exactly are we multiplying if we are awarded an allocation? Is it exclusively endowments and other noble causes? Do we get to know?

@Veridian, this is a useless debate, and it should not be in this thread. It is too elementary, so econo 101.

Vanguard manages some $3T. The money does not belong to Vanguard, they only manage it for millions of people from all walks of life who want not only to preserve their capital but have it appreciate for their retirement. For the service, Vanguard charges a well deserved fee, less than 1%.

So, if Vanguard manages to get market average (r_m), the bulk of it goes to the people they provide the service to. Since historical r_m is close to 10% CAGR dividends included, let's make a rough estimate: $3B for Vanguard, and $297B for the people. That is a very noble cause.

The millions of people using Vanguard as their money manager do it in spite of us able to generate much higher returns based on our backtests (r_m + α). They must have a reason. I think that the primary one is: they know they will win.

The millions of people putting their confidence in Vanguard did not rob or deprived anybody else. Market returns go to those participating in the game, those that take the risk.

If you sell your shares, the other side did not lose, it just started its participation. Lotteries are won by those buying tickets. If you do not buy any, you can not claim a prize, but you can say, in probability, that you did not lose a dime either.

Been actively listening to this post in hopes of a change of heart. I completely agree with the disappointment felt by the community. I came here specifically because of Robinhood integration with zero knowledge of either Python or Algorithmic investing. I'm an engineer by trade and found it to be a tremendous opportunity to learn Python, learn algorithmic trading and strategies, and have the ability to earn money while doing so. To that I'm thankful for Quantopian, but I'll move on once the first place that opens up Robinhood integration is set up. I don't have a chance at winning the lottery by getting one of my algorithms selected, which is the same fate for 99% of the community.

The obvious problem here is that core functionality for the community was abruptly taken away without anything to take its place. If there is actually a path for your average Joe user to contribute incrementally to the fund effort then it could have been launched concurrently with this announcement. Option A was taken out of play without a tangible Option B.

This has been a terrific thread. Just starting here, and it really gives me an insightful overview. So here goes: my interest is not in making money so much as it is in the very long haul. I believe that ai, robots, and automation will take over the world economy in the next fifty years or so. Some sort of ai will be providing data on companies (broadly speaking) that will profit from whatever the state of the world is by then. Selecting those companies will be the real market task by then to pick market return for national funds that provide everyone with an appropriate standard of living. Imagine, for instance, that this national fund will provide $10 T income for everyone in the US. If the market return is, say, 10% to make the arithmetic simple, that means a national fund of $100T. Figuring out how to do that without corruption is our real goal. Seems to me that Q is the way to go, and if those who are just using it as a convenient trading vehicle for personal gain want to scurry away, good riddance.

@Grant, agreed.

Individually, we are dealing with different perceptions of the same datasets trying in each of our own ways to arbitrage the quasi-random like price movements based on whatever criteria we may fancy at any one time or other.

Nonetheless, the goal, for anyone of us, is not necessarily to win the lottery ticket of an allocation. I see it more in designing the highest alpha generating strategy at least risk possible or acceptable. This even outside of Quantopian's preferences.

Over the long haul, it is the alpha that will make a difference. No one, at the end of the game is going to ask: were you scared? They will only want to know how well you did compared to a benchmark's metrics. If they also ask: was it tough? Answer yes.

Regardless, we do have to make choices. The question is: is there enough left under the Quantopian sun to still justify using it?

It is like in any other type of problem. You use the tools at your disposal until you find better ones, or find the ones that can do the job you want. I don't think we are here to answer Quantopian's needs, but our own. Surprisingly, at times, those objectives or preferences might coalesce.

Dear Q team,
I want to thank you for you countless hours of developing this amazing free service. I recognize that you have changing goals, want to reinforce others and have to put food on the table or cash in. Weather your decision was guided by one or all of the previous motivations I still want to commend you for what you have accomplished thus far. Balancing the constraints of a demanding, non-paying community and corporate funding must be a difficult feat I am sure. The fact that you have provided so much free content thus far regardless of your motives is still worth celebrating because we have, as a community, benefited greatly. Speaking for myself, through this community my knowledge of the investing world has increased substantially. So I just wanted to thank you again and want to encourage you on: onto many more years of bold ideas, pushing what is possible, and sharing many more successes with a great community. All the best.

There are WAY too many posts on here that I agree with, and I wish I could list them all, but a few points I think should (need) be made:

Dan Dunn says that there wasn't evidence of algos that would fit the institutional model. Why on earth are we going to spend the time optimizing an algo for a $1mil portfolio when the majority of us have $50k to trade and the possibility of getting an allocation is pretty much zero? Furthermore, developing that sort of strat is pretty hard to do when we can only trade equities, and futures for live trading just became a thing how long ago?? Creating institutional caliber strategies when we aren't provided institutional caliber data is a fool's errand. I am very grateful for Morningstar and 8,000 equities, but c'mon, we all know that's just a sliver of what's out there. Uncorrelated strategies at HFs are not using the 'Q1500' . I think a good example of an institutional strategy is risk parity, consisting of foreign/domestic govt bonds, TIPS, currencies, options, futures, equities etc etc. We had the option to use two of these asset classes, with US only constraints. If you're wondering why you didn't get enough serious algos, I personally think because at the end of the day, we all knew we'd look somewhat similar considering what we had to work with.

Its also very disheartening to hear that Quantopian wants to cater to the institutional crowd, as if every other HF isn't already doing this. Fawcett's post makes it sound like the institutions are bleeding out and they need you. That's sort of pathetic. When their AUM is in the billions, I seriously doubt they're going to come crying to Quantopian for help. The so called quant revolution has been happening for years. The picture it seems you're trying paint is that it hasn't, and you'll be the first to capture it. You guys are great at what you do, but you're dreaming if you think the CALPERS of the world are coming to you instead of hiring a handful of brains (like some of the members on this site) to bypass you completely.

Your platform gave us "little" traders power. You gave us the ability to feel like we could compete with the Bridgewaters and AQRs of the world even if we only had a fraction of AUM. THIS was your value proposition. The real reward wasn't winning your competition, it was creating something for ourselves and profiting from it. I'm sure I speak for a large portion of members on here when I say I would've been more than happy to pay a monthly fee to keep the status quo. You should know first hand how much us traders are willing to shell out in order to create portfolios/monitor performance. You could've taken that money and built out more of your infrastructure, supplied more data etc etc. While you may have had long discussions on this, it doesn't seem like you've thought long about the ramifications of ditching the little guys.

You got greedy... As if there wasn't enough of that already in the world of finance.

Out of respect for capitalism, I wish Quantopian the best of luck. But more importantly, I truly believe you're going to need it.

@Tom . You make some very good points. I think Q's real value proposition was to compete with the Wealthfront's of this world, not the AQR's.
The fact that Q has taken such a radical solution seems a bit surprising. There would be so many simple solutions that would not strand thousand of small time players and wannabe quants, whicle continuing to provide a robust community environment where future quant stars could emerge.

We know the Q solution suffers from the 1 minute bar restriction - which effectively discards it for many intraday trading applications. But that still leaves throusands of other profitable applications to be developeed that could benefit live traders, without necessarily fitting into the Quantopian risk profile of $1 mm plus account sizes and low beta rules.

If Q no longer wants to spend money supporting IB and RH, but only have to maintain the interface to the one broker it has chosen to deal with - and assuming that prime broker does not want to deal with small accounts - why not simply provide an xml based uri messaging service that simply prints out orders as the algo's determine that a trading order is called for?

It would be up to the individual user to tie that xml signal into the TWS or RH api to generate a trade there.

Quantopian could charge a fee for each message sent, thereby generating millions of dollars of new revenue WITHOUT signifciant admininstrative costs.. This would make Q more attractive than ever, opening it up to users of many other brokerages.

Simulataneously, this would not endanger or infringe on Q's propietary data feed agremeents with third party vendors, as this data could not be exported through that XML. Only the buy / sell orders could be transmitted.

Quantopianers, what do you say?

@Tom, right on.

@Serge, bravo, yes.

For the small trader, a simple trade messaging system based on their programs could be sufficient. And it would open up the door to trade coping services managed by the small trader for their benefit, and their subscribers. Hope it gets done as it is elsewhere.

Simulataneously, this would not endanger or infringe on Q's propietary data feed agreements with third party vendors, as this data could not be exported through that XML. Only the buy / sell orders could be transmitted.

This may be more complicated than one might think from a licensing standpoint. For example, rather than sending buy/sell orders, one could transmit price & volume data (or other data, as well). For example, a price of $10.57 could be sent as an order for 1057 shares. My guess is that any scheme that transmits licensed data directly into the hands of individuals will be a non-starter, given the efforts Q has put into keep the data with the confines of their API.

That said, if someone has the time, they could call Nanex and ask them what it would take to provide a Quantopian-compatible feed, and report back here. My bet, though, is that they won't have a solution, since any feed that can't be audited could be split into N feeds and sold, under-cutting their business. Quantopian has the advantage that the data usage (historical and live feeds) can be audited from a licensing standpoint (and Quantopian has a lot to lose by violating licensing terms, whereas all bets are off with individuals).

IBridgePy just posted an article on how to migrate your code from Quantopian:

http://www.ibridgepy.com/2017/08/27/tutorial-migrating-from-quantopian-to-ibridgepy/

Thought this would be of interest.

LOL--good bye Quantopian. Absolutely no reason to use this service now with QuantConnect and a hundred desktop applications that can perform backtesting.

I am deeply dissatisfied with the stopping of the live trading algorithm. The point of quantopian was to create an algorithm, refine it with the backtester and then use it to live trade (you can also get an allocation but no one will use it for that specific purpose). There is no point in using your backtester with another company's live trading software. Thus there is then no reason to use quantopian. I really hope you will change your idea. At most, make us pay for it, an extra commission or a small monthly fee will be fine with almost everyone, like other comments have mentioned.

At least make us pay for it. With a monthly fee (10s of $), everyone will be glad to pay it. Shutting down the live trading is a huge step towards the death of Quantopian.

I, as well as you, feel a strong frustration losing the possibility of handling my own investments in an automated way.

But you have to look forward, I do not think it is simple, but I think you can find a satisfactory alternative for this purpose.

Meanwhile, within Quantopian, you have to start putting yourself in the shoes of someone who has more than 50 million dollars and design strategies that work with that level of capital.

I understand that, regardless of the cost of maintaining the connection with IB, the Q team aims to eradicate the thought of strategies to manage personal savings, they expect the community to be fully focused on hedged strategies where the unit of investment is 10 million dollars.

There's no point William, they already said the amount of people actually live trading was so low that even if every one of them paid each month it wouldn't be worth the developer's salary and server costs.

So far (because I use pipeline) my only path forward is through Quantconnect and IB but I will have to pay $20/mo to live trade and $1 commissions on all my trades. Meaning that I will not be nearly as profitable as I have been on Quantopian but they are trying to get RH integration. The option that I am really hoping for is zipline live project. But they need more help and time to get pipeline working.

@Karl, agree. Mr. Buffett, as a student of Mr. Graham, has said the same thing for a long time. And, I agree too. Here is a small graph from my research:


What the chart says is this: over the life of a portfolio, stock weights will shift in favor of the highest alpha stocks, and be dominated by the best performers. They will be responsible for most of the overall return. And thereby, their sheer presence will not only act, but, become this weighing machine. For instance, you will find that Mr. Buffett's 8 biggest holdings out of his 90+ something account for over 65% of his long term portfolio.

http://www.ibridgepy.com/2017/08/27/tutorial-migrating-from-quantopian-to-ibridgepy/
This tutorial is about migrating from Quantopian to IBridgePy.
IBridgePy was officially introduced by Interactive Brokers on Nov 10th 2016 in their webinar.
The webinar is published by IB at youtube https://www.youtube.com/watch?v=hogXB07OJ_I
A lot of Quantopian algorithms can run on IBridgePy even without any changes.
Disclaimer:
This is Dr. Hui Liu, the developer of IBridgePy.

@Karl, the chart is simple.

Vertical axis: portfolio weights (20 stocks). All stocks start with an equal weight (5%) and are ordered by return (r_m =10% ± 1% alpha separation). Half will rise with positive alpha, the others will decline over the long term.

Horizontal axis: years (40). As time progresses, the higher alpha stocks take more and more space within the portfolio, their relative weights increases: w(i) / Σw, with Σw(i) = 1.
They grow faster than the others to such an extent that only a few of the stocks (the highest alpha generators of the group) will represent the bulk of portfolio return. The best 4 stocks gave 90% of the long term profits!

In my opinion IBridgePy looks pretty amateurish compared to Quantopian's platform. IBridgePy did an IB webinar last year which I attended; I didn't get the impression that the people behind it are professional software developers.

QuantConnect looks more professional and is outwardly most similar to Quantopian. My issue with QuantConnect is that I agree with someone else who pointed out that it's not Python-based; they have Python bindings but it's just C# in Python clothing. Not ideal if like me you want to stick with the Python data science stack.

I also came across QuantRocket which says it will support Zipline along with another backtester, with live trading and data from IB. Doesn't help me right now because it's still under development and says it won't launch until early 2018. Also, they have a free tier but the paid tier is quite a bit more expensive than QuantConnect at $100/mo and up.

I've been searching for an alternative these days. But honestly speaking, none of them can compared with QuantOpian, in my eyes at least. I miss QuantOpian a lot. But this doesn't help. I have to keep looking for another one. Time is money. That's the life. :-)

@Karl, the question is almost irrelevant having no real predictability at that level. What would be the alpha on a particular stock in 40 years? I certainly can not answer that. Well, yes I can: I do not know. It might even be nonexistent to outright negative pushing a stock to extinction.
Evidently, it should be the highest possible alpha, even higher than what was presented. This would apply to a portfolio of strategies as well; as I illustrated in the Alpha Vertex thread: https://www.quantopian.com/posts/alpha-vertex-precog-dataset

The maximum alpha on the above chart was 0.10 making it a maximum 20% CAGR. Close to Mr. Buffett's long term CAGR. The range is from 20% down to 1% CAGR in steps of 1%. In real life, these lines would be jagged, but the end points would still hold. There is no lost in generalities by having smoother lines.

Traders don't seem to consider such a chart, but they should. Even in trading the same phenomenon occurs. Traders could compensate, even accentuate this kind of development. Notice that the chart does not say continuous rebalancing to equal weights. On the contrary, it says let your high fliers shine, let them take in all the weight they can.

However, I hardly see any presented code on Q even trying to account for it. They will trade stocks like AMZN, AAPL,..., make a few bucks out of them, but have no provision for keeping them in their portfolios even though it was easy to do so. What I often see in strategies is throwing those few best performers away as if they had no use or need of them. Sure, trading strategies can not do everything, but... throwing away the best makes it that much harder to outshine with the rest.

That chart is something Mr. Buffett understood a long time ago. It is a side effect of having a portfolio of stocks each growing at their respective CAGR rates. The highest of the group will take the lion's share of the returns. We should therefore facilitate its growth and not curtail it.

@Guy, @Karl -- can you start your own thread instead of hijacking this one? This thread is about the brokerage integrations being discontinued.

Hi fawce, Dan,

Only had the time to comment on this now.

First and foremost, let me express my sincerest gratitude to Quantopian. You’ve opened the door (to me and many others) to the complex world of quant finance and trading, making it completely accessible to anyone and you’ve done it with incredible simplicity, sense of community and, let’s not forget, completely free. You achieved something very special!

Not being able to trade live with my own money is a serious setback, but still I look forward to continue contributing by learning, experimenting and participating. Most of the data, the research capability, backtesting, the tutorials, the lectures…etc., are still immensely valuable tools at our disposal and I believe that exploring them is the greatest investment one can make and that will yield the biggest returns. Thank you for keeping it accessible.

A note of worry: the option to alternatively trade your own money if you don’t get your algo allocated is arguably the biggest pull for newcomers that are willing to invest a lot of time researching and coding. My bet is that your numbers will suffer significantly, since there already are qualified alternatives; that ironically were following your footsteps…

A question I think you should have answered is: granted you have a few hundred members trading live, accounting for a small percentage of the community, but, how many of those are solid contributors to the institutional aspect of Quantopian in relation to the community? I don't believe they'll be motivated to stay. I suspect the opportunity to trade live with your own money is what's feeding true talent, in a way that you’re still motivated to evolve and develop since you can still make use of your work at your own risk if you don't get an allocation. With this measure, part of that motivation will dissipate...

Thank you again, truly!

Is the final take away from this thread that if you want to live trade you have to move to an alternative like:

as Q is not offering trade your self model and if you want to participate in the competition and / or get an allocation you should use Q?

As for other pursuing the allocation model we have:
- Q,
- https://info.cloudquant.com/,
- https://websim.worldquantchallenge.com/en/cms/wqc/websim/,
- https://www.quantiacs.com/Home.aspx

In pursuing the competition model we have:
- Q,
- https://websim.worldquantchallenge.com/en/cms/wqc/competition/,
- https://www.quantiacs.com/Home.aspx

Also it would be of interest what what others follow each model, i.e.,

  • trade your self
  • competition
  • allocation

And also what would be pros and cons of each and why is Q better? Initially Q covered all the bases which others did not do.

If you have your own infrastructure, you can shoot for allocations via FundSeeder, the various trading competitions, that contest/party in Miami...

Thanks for the above.

Also something I noticed is that about a days after this announcement the number of QC users reported on their website was 33k. Now it has gone up to 37.4k. So the option to live trading must matter to some like the competition and allocation even though there were few live traders?

Quantopian staff could at least leave some form of signals(https, email) so when the algo would trade so we could connect to a simple executor. Or execute manually if it's a slow algo.

There is http://www.zipline-live.io/ by Q which still does not have all the Q functionality and hope this is rolled out soon and also QunatConnect (https://www.quantconnect.com/forum/discussion/2400/avoiding-vendor-lock-in-running-lean-on-your-server) but I belive the issue will be the data and integrating with data feeds. Like Q and QC the back tested in QR (https://www.quantrocket.com/) might be open source. So there is an option to connect with a broker but I am not sure how practical it is to use.

So, unlike most people in this conversation, I only discovered Quantopian a couple weeks ago, and didn't have time to reap the benefits, and I am still massively disappointed. I did spent a lot of time learning Python and creating my first algo only to learn that I can't use it live. The thing I'm most disappointed about, though, is that there was absolutely no warning from Quantopian so that I could stop learning this and quit wasting my time the past couple weeks. Also leaves Robinhood users with no options, which is frustrating, now that I got a taste of what could have been. I guess I understand the variety of reasons to go this route with your business, but pulling the rug from under your users is mean.

You should really update docs and posts as soon as possible, otherwise other people will only sign up thinking robinhood integration is possible, like me. It was very hard to find information about broker support being deprecated until I finally found this post. Might have to find something other than "Leveling Wall Street's Playing Field" now.

@John Fawcett: From the outside, it's hard to see why providing a service to institutions is mutually exclusive with providing services to individual investors, especially if you left the current functionality untouched for individual investors.

Though still not viable due to data feed connections and lack of pipeline, Q is working in an alternative: http://www.zipline-live.io/. Also there are other vendors who provide live trading: https://www.quantconnect.com/, https://www.quantrocket.com/. All these support Python hence you skill is not wasted. I believe they want the live traders to switchover to the competition because this segment is not profitable.

@Suminda Thanks so much for the suggestion. After a night to think about it, the problem is definitely not Robinhood, it's the lack of Pipeline, or similar data. I still feel like Quantopian could have left the tools and structure for individual investors, with only the community and no support, and still pursued the institutional investor with their current resources. Still very disappointed.

A year of waste. As a 25 yo new to the community and quant trading this is a major disappointed. When I first found out about Quantopian after joining RH at the beginning of the year I was so fascinated by the Idea of being able to develop an automated trading system. What drew me even more to Quantopian was the contest. The thought of being to build something and then get paid for it if it was good and seemed like a great concept. Also knowing that there was and escape goat to trade that same code for your own account if it wasn't chosen pretty much sealed the deal. I knew I was a long way from being at a level needed to develop a high end and efficient algo that would make the cut, but as the saying goes practice makes perfect and you need to start somewhere.

Pretty much came to Quantopian with this exact thought process.

1) Teach myself python,
2) learn what makes a good strategy through the references and forums.
3) code an algo. and enter the contest. If didn't win but still made returns trade on my own account. Eventually if I could have made consistent income, I would have quit my full time job and focused full time on research and coding instead of just a couple hours a week in order to hopefully develop the large scale algos that Quantopian was looking.

Like all hedge funds you need full time coders and researcher and I'm pretty sure none of them are working for free. In this case the live trading seemed to be the income source of some of the individuals on and with out that you are basically forcing your opened sourced employees to leave.

It would be my luck that I get 95% though my first algo and this happens before it ever gets to see live trading. I guess it time to just look for a job at a hedge fund and get a steady income and a big bonus for doing well.

So now quantopian wants its users to work hard and get no returns. What will motivate people to write code if they cannot implement it. Also finding other alternatives is not a solution. Now we got to spend even more time learning another platform and most likely there will be limitations on the brokers there too. I cant believe I wasted my semester doing your tutorials. You throw out the alternatives as if everybody knows coding well enough to integrate their broker by themselves. The truth is quantopian got a couple of good algorithms and the profits are better than running the website like they did. They never cared a single bit about the community. They just wanted free algorithms from the users. How are you "Leveling The Playing Field" now? What about the hundreds of users that are halfway finished with an algo they worked on for a year and they can never trade it? where is the incentive there? The hopes that it will win a contest and get selected for funding? So if someone is smart enough to write an algo that great but not amazing he cant benefit from it. Im cancelling as will thousands of others. What a huge lying disappointment.

What I really want from Quantopian is: Adding a message system based on their Paper Trade, when Paper Trade trigger a trade signal, sending a message, via calling a Restful API which URI is defined by clients. By this clients can do real-money&real-time trade by their own brokers. The format of message is defined by Quantopian, supporting trade type, number and stocks etc., like what order_target_percent is doing. Quantopian could charge us for the cost per message.

As an alternative solution of shutdown current brokerage integration, the modification effort of this is minor and also the maintain cost of integration with brokers are transferred from Quantopian to clients. General speaking, Quantopian get their purpose and clients are also not pain, win-Win solution.

Adding a message system based on their Paper Trade...
Are you sure the paper trading is realtime? If not, what you get has delay. Sure if your algo is base on daily charts, there is no problem.

But I've asked the same question formerly but the answer is NO.

https://www.quantrocket.com/ seams to be trying to be doing just that. You can plug in your back tester. Apparently they will be supporting Zipline also. Since this is not yet released, do not know what data it will have or if there will be quakes. Also may be a area Q and QR can collaborate since Q is working on http://www.zipline-live.io/

@Thomas, Paper Trade is about 15 minutes delay, am I right?

If it's only 15 minutes delay, that's ok for me. It's not acceptable for more than one day. I think if Quantopian want to open this business, they are able to make it in real-time, since contest is based on real-time.

@Suminda, I see on their frontpage: QuantRocket is currently under development.

It's a pity if Quantopian not to do so, since they already have the resource and community, also this business may extend bigger and bigger, and mostly importantly, by this way they can have the strategy running on their platform.

I'd stopped "listening" to this forum thread, but there is an interesting topic here. I've been involved with Quantopian (Q) as a lowly user since 2012. It has been remarkable, the amount of money poured into the effort. Fawce & Co. must either have friends in high places or have a mighty fine elevator speech. In any case, there was a grand experiment of offering sophisticated retail trading to the masses, and it basically failed. Out of more than 100,000 reported users with some pretty darn-good resources, we reportedly have ~ 200 ambitious folks (and perhaps some small start-up "institutions") trading real money. I'd think that anyone considering starting up a business in this domain would need first to explain why they will succeed, when the Q experiment with retail trading failed. It sure seems like the market is not very well understood. Why would the various traditional "online brokers" not have failed miserably by now, or at least seen the writing on the wall and started to play catch-up? At one point, Q was working on partnering with E-Trade, and the deal fell through and Robinhood jumped in. The market is saying something when you can't even give away your product. Any insights?

@Grant, I haven't been around here since 2012 so I don't know what was Quantopian plan at that time. Was it the Q hedge fund or was it the retail trading one? If the main Quantopian goal has always been the hedge fund then the failure of retail trading business is less dramatic. Quantopian didn't invest all its resources in expanding the retail trading features (e.g. how long people have been asking for FOREX?), this might explain the failure.

It would be interesting to know how many users are actually live trading their algorithms on QuantConnect (37000+ users), a platform whose core business is providing live trading features to their users. This would give us more insights on the state of algorithmic trading at retail level. Unfortunately, we don't know those numbers.

@Grant, I believe the reported 100K+ users refers to registered users, not active users. I suspect the number of active users is substantially lower. Last year when Quantopian published their paper on in-sample vs out-of-sample performance using Quantopian backtests, I remember being surprised that out of such a large user base (already over 100K at the time) they only found 888 backtests to include in their study. This suggests to me that a lot of users sign up, poke around, clone an algorithm or two, but are not active, serious users of the platform.

This may not materially change your point about how small the number of serious retail quants appears to be, but it does suggest that live trading per se may not be the point at which the drop-off occurs. When I line up the available data points -- 100K+ registered users, <1K usable backtests for the Q paper, and a few hundred live traders -- the number of live traders looks reasonable relative to the number of usable backtests; it's the number of usable backtests that looks surprisingly low relative to the registered user base.

@George, as you started talking about numbers what about contest entries? They are ~500 only. Considering that each user can submit 3 algorithms, then we have less than 100 users competing. This and your consideration should tell something about how many users have the interest and skills to write trading algorithms. As for the Q hedge funds a few dozens of algorithms are probably enough (I guess), the number of active users is not a problem, but it is for keeping live trading alive.

Over all its a net positive. Move on to other platforms and solutions no big deal. The main thing is that the retail algorithmic trading scene is really small. Thats a huge net positive for people like us, we can pick up edges the institutions cant touch and its not a crowded space amongst the retail guys. I am more elated about this moving forward. Zorro Trader looks like a simple solution as an alternative, Lite C (less intensive than writing trading logic with pyhton) Its free also, then there are many other solutions out there that work.

Well, the inability for Q to grab a chunk of what would appear to be a large market is rather surprising. Think about all of the venues for retail trading--Ameritrade, E-Trade, Scottrade, Merrill Lynch, Interactive Brokers, and on and on. What are all these people doing, and why was Q not a good fit for them?

The Q hedge fund concept was introduced in the fall of 2014 (see https://pando.com/2014/10/16/calling-all-quants-quantopian-is-launching-a-hedge-fund-and-wants-you-to-manage-the-money/). The introduction was fairly general, with a notion that if lots of uncorrelated algos could be cobbled together, the result would be spectacular. The Q paper (discussed here) perhaps was a wake up call that a more sophisticated approach would be required; even an infinite number of monkeys on typewriters will not produce a hedge fund. The latest insight, it would seem, is that the Venn diagram of individuals capable and interested in conjuring up algos for a hedge fund and individuals doing algorithmic retail trading with their own money is a diagram of two nearly non-intersecting circles. There isn't enough synergy for it to make sense to offer both. The arduous, uncertain slog to get to the point where one could queue up for a six month wait for possibly no feedback whatsoever on a Q fund effort and then another long wait for an even less certain payout is a very different equation compared to controlling one's own destiny with one's own capital (however meager). The Internet offers a great opportunity to employ Pavlovian conditioning on the masses, but if only a few are rewarded, then only a few will become conditioned. It will be interesting to see what Fawce means by "we need to reward more community members" and if rewards could be extended to all community members. The other thing that the Internet enables these days is micropayments. My guess is that the framework presented by Jonathan Larkin is a tried-and-true recipe from the industry; it is nothing new. The question is how could it be adapted so that an itsy-bitsy contribution (e.g. a weak, transient alpha factor) could be rewarded with an itsy-bitsy payment (e.g. cryptocurrency). It would be easier to get the crowd to work for peanuts than for no reward at all (and trying to get the crowd to pay to work, a la Tom Sawyer, by charging for data sets seems completely unrealistic, but I ain't no psychologist--why are non-free premium data sets still advertised?).

In a world where every price is known but the value of none inasmuch, a faith in the benefit of doubt is priceless.