Back to Community
How is quantopian doing?

Has any information been published on how many people have gotten allocations?

How much money has been allocated?

Returns on the money?

Profits paid to algos?

27 responses

Is there any news on this? I won one of Quantopian's 6-month contests and I cannot even find out if my algorithm will be reviewed for potential allocation. In the Community Forums, there is a lot of discussion about how to get an allocation, much less discussion about allocations actually having been made.

Have there been any discretionary allocations since 2016 (https://www.quantopian.com/posts/quantopians-first-discretionary-capital-allocations)?

Hi David,

Part of the trickiness of building an asset management business like ours is that we are subject to regulation relating to publication of results and how we market. We need to be very careful about what say and don't say in the public. Yesterday, we were just discussing that an update to the community is warranted at this point. We'll be working on one, but it might take a bit to draft and review.

In the meantime, if you'd like feedback on your algorithm, you're welcome to post a tear sheet akin to what Grant has done in another thread or send a tear sheet to the support team for private feedback.

Regards,
Josh

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.

Hi Josh -

What regulations are you referring to? And when you say "asset management business" are you just referring to the 1337 Street Fund, or Quantopian, or something else? A Google search coughs up SEC FORM D for the 1337 Street Fund. There, it says the fund is not registered as an investment company under the Investment Company Act of 1940. Additionally, it is set up in the Cayman Islands. There are other boxes checked, as well, that presumably might have bearing on what you can say and not say at any given time, and how you say it.

The other perspective is that while your asset management business may be set up to be able to say whatever you want, you may still be restricted by legal arrangements with investors in the business or folks putting up capital, like Point 72.

It is also reasonable that you may decide to withhold information, not due to legal restrictions, but as a matter of how you chose to run the business. I would just make a clear distinction.

As a legal matter, perhaps you could invite counsel, Derek Meisner, to chime in here (listed on https://www.quantopian.com/about), or to provide a short brief (which you could post permanently on your website, and just refer to it the next time such questions come up). Seems like a teachable moment for the crowd, so why not bring in an expert? He could follow up here, as well. I'm still not clear if you cannot offer the PPM, due to legal constraints, or if it is simply a business decision on your part. I guess the issue is that public offering of information could be construed as offering the fund to the public, but it would seem that if the fund is clearly described as a private one, and that it will be offered only to qualified investors, then it would not be misleading to provide information to the public. Or maybe you could only make the information available to registered Quantopian users; it would only be accessible by creating a Quantopian account, and logging into it, and therefore technically would not be public.

Regarding "Yesterday, we were just discussing that an update to the community is warranted at this point. We'll be working on one, but it might take a bit to draft and review" I would think of how to institutionalize your updates. Why not send out a quarterly report to the crowd, who is working to build your business? Frankly, I think you have an obligation, if you are expecting engagement in and commitment to the crowd-sourced effort. There might be an unanticipated pay off.

Cheers,

Grant

An encouraging word if I may, as an extraordinary algo I evolved to, only does well at levels far below $10M so I had to look for a place to trade that and the more I see of __ unnamed competitor __ the more I like Quantopian, faster, easier and so on. :) Appreciation.

I also won one of the monthly contests and as a result my other entries were removed from future contests... unfortunately because they were doing quite well. I actually think these algos were better than the one that won but I have no idea if they're being reviewed anymore for allocations. It would be great to get more feedback on reviewed algos that did well in contests.

Congratulations Jay Ross. Personally, at this point, I feel both discouraged and disenchanted. If winning algorithms regularly go without review, I have to wonder if all the hard work of creating and researching institutional grade investment strategies by the Quantopian community is actually worthwhile?

Grant Kiehne, I fully support your comments above. Thank you for sharing. In particular, I too think Quantopian has an obligation to provide much more transparency (if only by providing consistent feedback to authors privately) if they expect engagement in and commitment to the crowd-sourced effort. Otherwise, what is the point?

I also won third place for Contest 34 and 35, also currently in the the top ten of the new daily contest but haven't received any feedback from the Q team. I'm thinking that since Q changed the default slippage to 5 basis point, while it doesn't affect the results of OOS live simulated paper trades, it might have adverse consequences in their long backtest where they determine if the algo's performance will hold in different market conditions using this new default slippage. I might just be speculating here but if they go through this process, the proper way to analyze long backtest results is first figure out the "real" slippage based on OOS live simulated results and then use this in the long backtest rather than defaulting to the fixed 5 basis points which I think is quite conservative.

I can also understand why Q keeps changing the rules, requirements, thresholds, etc. in the middle of the game as they are getting real feedback from actual trading of their hedge fund and still honing on exactly what they are looking for in terms of their final product. Nonetheless, it would be nice for at least the top three placers to get some feedback regarding possibility of getting an allocation or give specific reasons why they are not receiving allocations with advice on how to improve the algo to make it feasible. After all they designed the contest to extract algos they will allocate funds to, so at the very least, they owe it to the winners to communicate their analysis of the algos instead of us wondering what is wrong with our winning algos!

I too noticed that default slippage has been changed to something more conservative/penal. Like you James, I am not able to replicate my live trading performance. Moreover, re-running backtests I had run in the past now generates very different performance.

@David Hall, I agree it can be discouraging and hence maybe why I've significantly cut down on developing new algos over the last few months. I feel like winners of the monthly contest or even the top 5-10 should be entitled to some form of feedback from an allocation review. If the feedback is constructive, we can then go back and adjust our algos in ways to improve our chances of getting an allocation. Full disclosure, I spent maybe 200+ hours last year developing algos for Quantopian, luckily it paid off by winning the contest but I still don't feel as motivated I can be to keep working towards the goal of getting an allocation if I can't get feedback on all the work I've put in so far to know if I am at least on the right track or not.

Also to be blunt, the transition to order optimization had devastating effects on my most successful strategies. Their seems to be a preference for allocating a % of portfolio rather than using absolute $ values for ordering... for many reasons this impacts some of the best strategies I had developed. I understand the rationale behind the new ordering system, but it should have been more accommodating.

@Jay Ross, I cut down to no further development. I spent a lot of time and ticked all the boxes with little to show for it. Winning a contest was never my motivation. In that I suspect I am not alone!

My development went to zero awhile ago as well.

Once a couple of months passed from hedge-fund launch and I started seeing some departures from Q management and less feedback, no news seemed to be bad news in that they can't get the thing profitable (not that any other manager can)

Hi David,

5 Basis Points Fixed Slippage and 10% volume limit is an assumed transaction cost that according to Q is based on studies of their actual trading experience. Actual Slippage of the algo in the live simulated paper trading environment can be very different from this new default settting and also cause discrepancies when you re-run the old backtests. This is where the problem lies. Now Q might be running longer backtests to further evaluate your algo under this new default slippage and performance is degraded substantially that renders it below Q thresholds. As every algo have different design and levels of order execution efficiency, this really comes down to how to accurately account for slippage in longer backtests for evaluation of possible allocations. If your algo's performance in simulated live trading has a high order execution efficiency (i.e. low slippage costs), then applying the new default slippage to longer backtests might not be accurate and be overly conservative. With $10M + in play, every hundred of a penny counts!

I have to say, it does become tiring trying to save Quantopian from itself. The fundamental problem, I believe, is an under-appreciation for the potential of collectivism in building a crowd-sourced hedge fund. If one has a look at https://www.quantopian.com/about, it kinda has the right feel to it, but in practice, the crowd is kept at arm's length from the business. My read is that the whole thing needs to be re-thunk, from the top down, so that it is not just a "head scratch" if the crowd needs to be updated on the business. The crowd actually needs to be part of the business, with a legal obligation to provide information. It would be both an ideological shift, and legal one. For example, there is something called a cooperative:

an autonomous association of persons united voluntarily to meet their common economic, social, and cultural needs and aspirations through a jointly-owned and democratically-controlled enterprise

Sounds about right for a truly crowd-sourced hedge fund. Quantopian is not even close to a cooperative, however, and there is no indication that it will change. Bummer.

Some technical notes:

Q published a study "How Accurate is Our Slippage Model: Comparing Real and Simulated Transaction Costs" (dated December 6, 2016):

https://blog.quantopian.com/accurate-slippage-model-comparing-real-simulated-transaction-costs/

Here is the announcement of the recent change to the model:

https://www.quantopian.com/posts/changes-coming-to-the-default-slippage-model

Some other references:

https://www.quantopian.com/lectures/introduction-to-volume-slippage-and-liquidity
https://www.quantopian.com/posts/quantopian-lecture-series-market-impact-models
https://www.quantopian.com/lectures/market-impact-models

A related topic is the unaccounted cost of shorting:

https://www.quantopian.com/posts/short-selling-in-backtester-time-for-improvement-1

Not really a simulation accuracy problem, but there's also an issue with reported Sharpe ratios (risk-free rate is unaccounted for, per the conventional definition):

https://www.quantopian.com/posts/risk-free-rate-on-quantopian

One gets the feeling that things are going from bad to worse when some of the best algo authors, as evidenced by being winners or top placers of the contests, being disenfranchised because of Q's lack of communication or responsiveness to them. Feeling left out to dry after all the hard work one has put in to achieve the goal that is put out there via the contests is the cause of disenfranchisement. The carrot in the stick is the possibility of fund allocation, while contest prize money is just teaser. At the very least, Q should reach out to the contest top performers and communicate their review as to why or why the algo qualifies for allocation, advice them on improvements, etc.

That is why I was a little bit suprise when two top managers of Q quickly responded to a post asking for feedback here party-algo-feedback-requested-please Yes, it looks like a pretty cool algo that fits Q's intended strategy but not tested live nor was a contest top placer. So, it begs the question, why does Q not communicate such feedback to some of the contest winners? Posting your algo notebook and asking for feedback, is this a better way to get Q's attention rather than performing well in the contests? I am again left scratching my head that it's now beginning to hurt!

@ James - I've decided to start offering a 15% gratuity to any Q employee who provides substantive feedback. : )

@Grant,

Hey, if that's what it takes, I'll willing to do the same:) But I highly doubt that it's allowed, I'll stick to offering bottomless beer or single malt, perhaps as an added bonus, burritos!

This is just an idea....
There are some other really interesting business models out there that could attract investors and algo developer as well. I found this model very interesting (it is base in Europe but it should work everywhere - see
wikifolio.com

Note to select English language, select International (beta). It is quite an interesting model (and transparent). Have read through ....

@Carlo G, Thanks for the info. We have something similar in the US ...Collective2

We are at the dawn of disruptive financial technologies, more should soon pop up and competition will bring efficiency!

@James: Thanks for the link.
There are some differences - the Austrian Model in wikifolio is actually selling each strategy as a fund - I can buy each strategy as a fund on any trader account (IB etc.), so I do not buy the strategy, but the actual fund - the trading is done by the "trader" (or algo provider). It is "user friendlier" for investors - no need to turn on any auto tarding etc.
Also, the incentive for algo provider is quite nice - they get paid a percentage if new highs are made - so there is an incentive for both - algo provider (which in actual fact beomes a fund manager) and investor - as he/she picks the strategy with full transparencies.
I really like the simplicity of it.

I agree new FinTech will come .....

Hi folks,

I wanted to circle back to Miles' original questions first and share what I can there:

Has any information been published on how many people have gotten allocations?
- to date we have made allocations to 23 individual community members from 9 countries (several individuals have received allocations for more than 1 algorithm).

How much money has been allocated?
- as of today we have made over $150MM in allocations to algorithms.

Returns on the money?
- due to the regulated nature of our asset management business we are not permitted to share (i.e. "market") our returns.

Profits paid to algos?
- authors to date have earned or accrued 10% royalty fees on the profits earned by their algorithms

Speaking to the rest of the thread, I hear loud and clear that if you aren't one of the 23 people we've tapped to date it can be frustrating and hard to tell if you are on the right track to getting an allocation from Quantopian.

There are a few ways we're working to try to tighten the feedback loop, most prominently is the re-worked contest which is now closely aligned with our selection process. If you find yourself consistently winning, ranking highly, or even just qualifying your algorithm to be ranked in the current contest - you are on the right track!

We have learned a ton in the first 12 months of running a crowd-sourced hedge fund - coincidentally today is the exact (!) 1 year anniversary of our launch. We do our best to share that knowledge with you in the form of improvements like our more conservative default slippage model, our factor risk model, our educational content, and new youtube tutorials.

It has been an inspiring and humbling experience to collaborate closely with the 23 community members who have licensed us their intellectual property. We appreciate their efforts and the confidence they've placed in us to be good stewards of their IP, just as we appreciate all of you who make up this community and contribute your work, your feedback, and your energy to making this platform unique.

Looking forward, we aspire to scale our selection process up and make many more allocations in the next 12 months. While we can't guarantee that we will be able to make an allocation to every good algorithm on the platform, we feel strongly that every person who takes advantage of the community, educational materials, software platform and data on Quantopian can derive benefit from their efforts.

Another source of feedback is the community. You, the community members, have become adept at analyzing tearsheets. Recently, a couple of productive threads have progressed with good feedback between community members with respect to the quality of their algorithms. We encourage you to share your tearsheets for similar feedback.

And as always - feel free to ping me directly and send me tearsheets to review. If I get enough emails, we will schedule another webinar where I provide feedback and answer questions live (similar to the last such webinar).

thanks, jess
[email protected]

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.

Thanks Jess -

Nice post! And congratulations on the fund launch and 1-year anniversary! I'd consider making the information you share more prominent. I had to search for it on the forum; at a minimum, you could mark this thread as "Interesting."

Some comments and questions:

  • It is worth noting that your average (un-leveraged?) allocation is $6.5M, which is right in line with the contest capital of $10M. So, we aren't testing a hypothetical level of capital.
  • When you say "we have made over $150MM in allocations to algorithms" does it include leverage? I recall a comment by Fawce awhile back that you intend to use up to 6X leverage. It has also been communicated that licensees are paid on the leveraged capital. So, potentially, we could be talking about 6X$150MM = $900MM, right? Or at the other extreme, your $150MM figure could already include 6X leverage, and so only $25MM in capital was deployed?
  • Regarding "we aspire to scale our selection process up and make many more allocations in the next 12 months" it would be interesting to hear how this will work. Presumably, by "many more" you mean relative to the current number. So, in the next year, you'd make >> 23 allocations? And maintain the $6.5MM average allocation? Are these correct assumptions?
  • Eventually, one has to figure that the fund will get to a point where more incremental drop/add of alpha will be the norm, and you will apply the nice framework provided by your former colleague, Jonathan Larkin (see https://blog.quantopian.com/a-professional-quant-equity-workflow/). It would be interesting to hear more about how you intend to approach this, both at the individual author level, and fund level. Personally, I've tried to set up an algo architecture that supports multiple Pipeline factors, and combines them (presently, naively, with a sum of z-scores). So, in theory, I should be able to have a single algo, and manage it over time, with drop/add's of Pipeline-based alpha factors. What is not clear is how you will manage this, going forward. For example, on https://www.quantopian.com/posts/contest-algo-feedback, Delaney advised researching additional factors. However, it was unclear how this would work vis-a-vis the allocation process. I think it is a good idea to focus on researching individual alpha factors, that are expressed in Pipeline, but how this part of the workflow relates to the allocation process is unclear.
  • It would be interesting to hear about how you might apply the quant workflow to the fund, versus a simple weighted sum of licensed algos. I can understand that you would want to keep your fund construction and management private (and perhaps there are legal constraints, as well). However, I gathered that Jonathan adapted the architecture from how a fund would be managed, with each alpha factor provided by a quant, and the rest left up to the fund manager. Presumably, at this point, you are running each algo independently, with its own little pot of capital, and just summing up the returns across all algos. I bring this up, because it potentially impacts a number of areas, including scalability and how you compensate quants. It would be an interesting community discussion point, if you are willing.
  • When you introduced the fund concept a number of years ago, there was a strong emphasis on combining N uncorrelated return streams. Presumably, this is still the underlying theme. So, would it make sense to provide some automated feedback on the correlation of an algo to what you already have in the fund? I could write what appears to be a wonderful algo, wait 6 months for out-of-sample data, only to find out that it is correlated with what you've already funded, and is worthless. On https://www.quantopian.com/allocation, you have a "Low Correlation to Peers" (which makes a lot of sense) but there is no way to test algos against the requirement, as far as I know.
  • To what extent are the algos you currently have in the fund uncorrelated?
  • Presumably, one path to writing uncorrelated algos would be access to more free alternative data sets. Is there a path to provide more of them to the masses?
  • Over the next couple years, what is the timeline to add other types of algos, beyond the current focus on U.S. equity, long-short? I ask because at some point, I can imagine you'll have your fill of the current flavor, so if I'm to do research and it will take awhile, I'd like it to be worthwhile.

Hi @Jess,

Thank you for this update - very much appreciated!

If possible, it would be great to get these type of updates on a more regular basis (monthly if possible, or at least quarterly perhaps?), rather than just ad hoc.

We'll always appreciate any update you're able to share, but in particular, I'd be interested in the following metrics:

  • # of strategies deployed since last update & in total
  • # of strategies withdrawn since last update & in total
  • % of total, and/or absolute amount of investors' capital deployed (e.g. $150MM out of $250MM from Point72?)
  • % of total, and absolute amount of leverage deployed
  • Range of capital and range of leverage deployed
  • Any type of meaningful metric of correlation of strategy return streams
  • Challenges/lessons learned that you think may be useful for the Q Community to be aware of
  • Plan / targets for the upcoming allocation period (until next update) that you're willing/able to share

"What gets measured, gets managed" as I believe Peter Drucker once wrote, but I can understand if it's too premature to share some of this (perhaps also in order to manage our expectations).

Again, thank you for the updates! It's always very much appreciated. Personally I feel it's a very exciting period to be part of the Q Community!

Best,
Joakim

Thanks @Jess for your response and guidance. Congratulations Q on your first anniversary as a full fledged hedge fund! +1 on improving feedback loop and communications / guidance . It can only get better!

Hi Jess -

It is impressive that you have been able to achieve global reach ("community members from 9 countries"). Is this a kind of objective that influences the allocation process, or just a happy coincidence?

Grant, I would describe the fact that we have made allocations to community members in 9 countries as unsurprising given the geographic distribution of our community, rather than a coincidence. With community members in 194 countries, it is more like a likely outcome!

As suggested, I submitted my (contest 34 winning) strategy privately on May 18th for feedback. After more than a month without any response whatsoever, I'm officially a non-believer.

David,

I'm sorry we haven't gotten back to you with feedback. Did you send it to me directly? Please point me to (ideally attach) your tearsheet and I'd be happy to give you feedback on your strategy.

Best wishes, Jess
[email protected]