Quantopian and QuantConnect

As an observer of both the Quantopian and QuantConnect backtesting and algorithmic trading platforms, it’s interesting to see how they’re gradually coming to resemble each other. Quantopian, of course, is backed by GETCO and Spark Capital and is implementing new features rapidly, so it would appear to have the upper hand in financing.  It also has a partnership with Upgrade Capital to discover emerging talent.  Meanwhile, QuantConnect seems to be just now raising funds, but is starting to get noticed, and has a partnership with battle-fin.com to unearth young guns. Another major difference is Quantopian’s use of the Python programming language in its IDE, and QuantConnect’s use of C#. That’s good for broad market reach, as each language has its own following (for instance, my own programming skills are primitive, but I have some familiarity with C# and no knowledge of Python). On the other hand, perhaps each site could offer a lower tier subscription with a language-independent REST API to allow traders to send signals for routing and execution. The backtesting functionality is nice, but I believe there is a whole market out there for a webservice to trade with Interactive Brokers and cut the cord to the desktop. This would simultaneously address the concerns of those who aren’t ready to host the source code of their strategies on Quantopian’s or QuantConnect’s servers.

They both plan on executing through Interactive Brokers, which makes sense given its reasonable prices and broad retail market reach. Covestor‘s done it, but in speaking with the managers of several autotrading sites, Interactive Brokers is not an easy partner to deal with.

Best of luck to both companies. I look forward to writing a review of each when time permits. Provided I get around to learning some Python first.

Update 5/16/13: Forexthink.com also has a nice overview of QuantConnect.  I am especially interested in this part:

Looking further ahead, the firm has plans to set up a hedge fund based around the most successful managers, marketing individual strategies to retail investors via online brokerages, and packaging the best algorithms as an exchange-traded fund.

I could be wrong, but I seem to remember in a previous iteration of QuantConnect, they intended to serve as a sort of exchange to connect aspiring quants with investors looking for alternative investment strategies (thus.. QuantConnect).  It seems that they have not given up on that goal long-term, but are working on the basic technology implementation for now.  Interestingly, this also now puts them in the same arena as Currensee, which has set up its own fund of leading traders.  Let’s see if Quantopian follows suit now that it’s enabled live trading on its platform.

11 thoughts on “Quantopian and QuantConnect”

  1. It seems scary to me that Quantopian is funded by this big algo trading companies. What if this is just a big plot to steal peoples ideas.

    1. While I hope that’s not the case, given the reputational damage that Spark Capital and GETCO would suffer if such a thing were discovered, you’re not alone in your doubts. Many have expressed similar concerns about data security on Quantopian’s forums. That’s why I think a clever compromise would be to create a lower tier service that would just provide a cloud-based trade API to Interactive Brokers. The trader would keep his code on his own computer/server/cloud platform and send trading signals to Quantopian to execute at Interactive Brokers. Obviously, this would not work well for higher-frequency strategies, but I think there is a market out there that wants to remove the unpredictability that TWS/IB Gateway currently represent. Quantopian (or QuantConnect) could fill this niche with a web-service trading API (preferably RESTful so it’s language agnostic).

        1. At the time I wrote that comment, I had trouble keeping the IB Gateway constantly connected–it would drop connection at random times, sometimes aborting critical trades. It’s been far more reliable for me lately, but I still don’t like introducing yet another component that can fail (some desktop/server -> algorithm -> IB Gateway -> internet connection -> IB).

  2. Thanks for the article INTJ. QuantConnect.com wants to provide a way to backtest & trade given the extensive barriers to quality backtesting today. You can share your strategy results to gain kudos for your work. Potentially you could use Collective2 API for a live track-record tracking service?

    @QuantDude – We shared that concern and intentionally steered clear of any potentially compromising capital. Its been a harder road but we’re glad our motives remain transparent.

    1. Jared, I already use Collective2, but C2 does not have many broker relationships, so I still have to automate my trades on my own computer. Not the end of the world, but IB Gateway is far less reliable than I hoped. A cloud-based solution to trade would be nice.

      1. Sorry misunderstood – yes a REST webservice API would definitely help reduce dependence on IB Gateway. We’re working on ways to keep your raw code local (e.g. only uploading a compiled DLL), so I’ll see what we can do to to help this angle as well.

  3. @INTJ, it might not address the type of reliability you find lacking in IB Gateway, but my product Brokertron Gateway for IB
    makes possible connecting to IB from your own server and addresses some of the shortcomings of IB Gateway.

    Specifically, Brokertron Gateway for IB makes possible IB login with a IB security device via a web browser instead of via a remote desktop or VNC connection to a GUI desktop and the IB Gateway GUI application.

    In addition to a web browser UI there also is a Web API for IB login too. However the IB API port that becomes available after login is still the usual language agnostic sockets based API from IB and not a modern REST web services type of API. You use the usual libraries from IB to connect to the usual IB API port that Brokertron Gateway for IB will open up.

Leave a Reply