Quantopian - use and abuse

I have been absent for a while and was wondering briefly whether to change the habit of a lifetime and do some short term automated trading. Its a pity that Q has abandoned this facility and I have been looking at Quantconnect who now accommodate those who like to code in Python.

QC is not nearly so active as Q and it does not seem as good a place to play and exchange ideas.

I have no idea what Q's plans are and no interest in its contests - low beta, low vol long short equity is not for me.

But looking around at the alternatives does make me realise once again what an outstanding effort these guys at Q have made. I don't happen to like running algos on-line since I like to drill down far deeper than is possible here but if you want to test on an intra-day basis this is still a pretty fine place to do so.

Must do some more fiddling around here. There is some genuine talent as well as the occasional nutcase.

5 responses

I think there are alot of people on the QC forums like myself that are lurking but there hasn't been anything engaging enough to jump aboard yet. I'm a big fan of the slack channel though, very helpful. The thing I miss most about Quantopian is the QS500 & QS1500 universes (a real shame you can use them in live trading anymore) as well as their research environment for testing factors & portfolio construction. Ideally QC would manage the live trading component but leverage the open source tools that Quantopian provides.

I haven't managed (or perhaps bothered) to work out slack: you seem to have to have a paid subscription? Also I am very puzzled by “Broadcast”. Only one system there? For python at least? Had not realised there is no S&P 500 universe.

It all seems very inactive and Jared and his mates seem pretty slow to answer questions. Probably setting up Lean or Zipline in the clud is the best alternative and buying your own data? But is it worth it? Last time I tried with Zipline I gave up in frustration - perhaps it's now improved and uses docker or at least some better method of deployment.

Frankly, looking at all the track records of all the hedge funds and CTAs out there over the years I wonder whether the whole enterprise is worth it anyway. It is well worth it if you are a manager of course - your big profits come through management fees.

But as an individual trader? Well there you either have to shoot the lights out on a small amount of capital and take the risk of blow up or deploy a large amount of capital and settle for lower vol, lower returns.

And let's face it, given the huge, overwhelming number of professional and amateur quants and coders out there these days, is much left to be discovered? I think not.
I'm a strong believer in AI in the long term
https://anthonyfjgarner.net/2017/12/12/can-artificial-intelligence-dream-of-electric-sheepdaviddeutschoxf/
But even then, even if we had general AI, would it necessarily have any better method of prediction? I don't know.

For me it all boils down to one simple observation: outperformance / performance is highly probably mean reverting. Here today gone tomorrow. And I do not believe any amount of clever maths or statistics will change that.

Hence in my view one probably wants to stick to a simple carry trade and hedge it to the best of one's ability with non correlated (ha ha) assets. Anything else is probably great for marketing but probably not much better over the very long term than a traditional 60/40 or simple risk parity approach.

Which is why I have re-named myself the financial grynch.

Call me a spoilsport but I have seen too much hype over the years and been in and out of too many hedge fund investments and strategies to believe otherwise.

Nothing wrong with Quant, nothing wrong with analysis. But if I was younger and keener and enjoyed people contact I would concentrate on building AUM in a low vol, relatively simple, unexciting fund and make my money that way. Now that IS a carry trade.

And good luck to Q: I wish I had the energy and people skills to tag along.

Slack is free & Just a live chat platform where you can ask questions. I don't use broadcast.
I have found the forum members & Jared quite helpful so far. Keep in mind they are constantly building new features. I don't think it's worth setting up your own execution + data etc... it far exceeds the costs provided by QC, don't forget who is going to support it when there is a technical issue?
We are in a new era where retail traders can access quantitative technologies without losing an arm & a leg and can focus on developing & deploying their strategies. Don't forget that this is all quite new, I am also frustrated with the challenges of QC every day but it is early days & I'm learning every day with how the technology works (Ray Dalio says life is about struggling well). Each algo issue is a learning opportunity. Institutional returns suck yes but there are many simple strategies that retail traders can now leverage without paying the performance fees & you can build portfolios to match your investment objectives. Being a retail trader is a lot better because you don't have to worry about investor relations, sales, account management, execution challenges etc...
The big benefit of QC at the moment is that they also support trading in futures, currencies, crypto & options allowing you to diversify your portfolio a lot more. Many profitable strategies are not available to institutional managers because of their size/market impact. I'm personally aiming for a portfolio of a Sharpe ~2 making 20% cagr & I think it's realistic (albeit challenging).
In summary it's still early days, QC will keep building new features & tools over the coming months to meet the demand from its clients & standards set by QT.

I think both Q and QC are excellent. To be given access to (vastly expensive) minute level data and a superb back-testing engine is worth a great deal. The tutorials provided by each are very valuable and stimulating. I admire what both platforms have done.

My audited track record for futures trading showed 16% CAGR before MF Global blew up so yes, I agree these sort of figures are attainable. But you have to have considerable capital to live off 16% after taxes. Most people here are trying to shoot the lights out for huge returns on small capital and I have no qualms with that - good luck to them. Although they would probably have more luck blindly subscribing for a few ICOs these days as I did in the 90s with tech IPOs.

My belief is simple: the big money in this game goes to those who raise AUM or who are lucky enough to ride the coat tails of an insane boom like we are seeing with cryptos and who exit before the inevitable upset. Flipping is probably the best bet.

I agree with you there, you can't live off the money unless you move to Mozambique but you can make some good side money with things like selling indicators/strategies/managed accounts etc whilst growing your wealth over the long term through compound interest & retire early. If you start with $100k, put in$20k a year & compound that at 20% for 20 years you will end up with \$7,567,520. Then if you're confident with your strategies then you can add leverage. I guess the question is, if you weren't doing quant trading, what else would you be doing?

Compound Interest Calculator: