Quantitative trading isn’t a new idea by any stretch of imagination. Searching for the Holy Grail in quantitative trading continues to fascinate both financial professionals and all sorts of scientists (physicists, engineers, mathematicians, etc). Data scientists are used to handling data sets and modeling them, knowing that their prediction model WILL NOT affect the hypothesis or the phenomenon that they are modeling. However, this is NOT true in financial markets.
I see four types of challenges in quantitative trading in financial markets beckoning solutions that have existed forever:
- The amount of Subconscious biases amongst Financial professionals developing and implementing quantitative strategies, is significant.
- Acting as adversaries on a Survival TV show, is the modus operando?? amongst Financial professionals.
- Allocating capital towards quantitative strategies is a process dominated by subconscious biases.
- Network effects has had no place in developing superior quantitative trading strategies-products.
As you already sense, I am focused on the human-related issues of quantitative strategies-trading. Don’t forget that even if we end up overweighting quantitative strategies in the allocation of our capital resources (as individuals and as institutions), we are still depending on humans that are coding, fine-tuning, adapting, etc these algorithms.
In this post, I will revisit Quantopian, a Boston-based Fintech that I covered in the very early days of my Fintech content creation journey. Check out Quantopian – DIY open-sourced trading algorithms and a crowd-sourced hedge fund. At the time, Quantopian had received around $23mil in funding (by March 2015). Today, they have another $25mil from Point72 Ventures, Bessemer Venture Partners, Andreessen Horowitz (source:VentureScanner). More importantly, they have partnered with Point72 Ventures who has promised to allocate up to $250mil to quantitative strategies managed by Quantopian.
At the lunch organized by the Swiss FinteCH association, the Director of Academia of Quantopian Delaney Mackenzie, said that Quantopian is now allocating to 17 strategies that are benefiting from the first batch of capital from Point72. This is crowd-sourcing quantitative strategies not crowdfunding, in action.
The lunch gave me the opportunity to understand the kind of people Quantopian is looking for, the kind of strategies they seek, and the way their business works.
Quantopian has currently, around 130,000 users from which they crowdsource strategies. This is a platform that can give the opportunity to non-professionals to validate a hypothesis that is coded into strategy. Quantopian is avoiding subconscious selection hiring biases, by giving the power of algorithm filtering to a machine. Quantopian has automated the process of creating a short-list of algorithms and authors-coders. Their platform at this stage is
Acting as a Discovery platform for Talent that is left unexploited for various reasons.
This may include academic staff that may have one-off brilliant strategies, data scientists that have no access to high quality financial data, young techies that are unbiased from the traditional financial modeling frameworks. From the current 130,000 user base, Quantopian has detected concertation of talent: (a) geographically, in 5 countries; (b) educationally, with tech related backgrounds; (c) in gender, men rather than women. The gender imbalance is actually an area that Quantopian, whose role is at the same time educational, wants to improve. They are currently, exploring various initiatives that can improve that split as they genuinely believe that that kind of diversity can improve the quality of the diversification they are looking for.
The kind of strategies that Quantopian is looking for, are Alpha-generating algorithms.
They have been concentrating in the US market due to the maturity and the availability of data for back-testing, live-testing and validating.
Quantopian offers a 10% payout to the “author-coder” that is selected and allocated capital. This is 10% of the net performance of the author’s algorithm; a rate probably above the average “bonus” of a proprietary trader but lower than most hedge funds (which come with other kinds of risks and responsibilities).
Quantopian is fairly open-source for the most part of the journey of its users. So, any user, for example, can build on top another user’s code (permission from the other user needed). Users own their own IP during the phase of coding, tweaking, and back-testing. Naturally, there is an inherent trust towards Quantopian who would not jeopardize its brand name, by stealing the IP from any user. This kind of relationship (based purely on trust and with the user-coder in the driving seat), comes to a halt once the code-algo is selected and ready for capital deployment. At that point, there is a boat-load of legal agreements that coders-users and Quantopian sign to license the IP to Quantopian. This is the point that Quantopian shares profitability of the strategy and reaps the benefits of creating an alpha-generating crowdsourced quantitative investment firm. What is important here to bear in mind, is that Quantopian has to keep up the game of “feeding algorithms” that continue to generate alpha, in a financial world that is affected by the trading patterns themselves.
Quantopian is addressing two out of the four challenges, I mentioned earlier. It is addressing both subconscious biases 1 and 4, traders themselves and those allocating capital to traders.
Stay tuned for a totally different approach to generating value using quantitative methods. Next Tuesday.
Efi Pylarinou is a Fintech thought-leader, consultant, and investor.
Get fresh daily insights from an amazing team of Fintech thought leaders around the world. Ride the Fintech wave by reading us daily in your email.