This post was originally featured on the Quantopian Blog and authored by Dr. Jess Stauth.
Last week I gave a quant finance meet-up talk at the Hacker Dojo in Mountain View, CA. The format was inspired by some analysis I did on the types of algorithms shared and cloned in the Quantopian community – initially I wanted to ask: What are the most popular strategies coded up on Quantopian? To answer this question I ranked all public forum posts three ways, first on number of replies, second on number of views, and third on number of times cloned. I averaged these scores and re-ranked the list to come up with the top 25 ‘Most popular posts of all time’. (NB: I did not do any correction for the date of the original post, so the amount of time the thread has been alive has not been normalized.)
Starting from this list, I worked backwards and used examples from the Quantopian community to introduce 5 basic quant strategy types: Mean Reversion, Momentum, Value, Sentiment and Seasonality. While this list is not technically ‘mutually exclusive and collectively exhaustive’, it covers a large fraction of intraday to lower frequency quant strategies and provides a good overview of the way equity focused quants think about predicting market prices. I went back to my Top 25 list and categorized each algo into one of these five buckets and then created this pie chart based on the aggregated number of views for each strategy type.
There are a number of interesting conclusions to be drawn from this initial overview of community activity. Perhaps the most obvious and predictable of these is that price based strategies are currently in the lead by a large margin – due, I expect, to the easy access to minute-level equity pricing and the accessibility of the logic for momentum and mean-reversion. Indeed there were no value-based strategies that made their way into the Top 25 – which in my view represents a key opportunity space right now.
More subtle and, from my admittedly biased point of view, more compelling is the diversity and quality of content and collaboration in the public sphere. Having joined the Quantopian team from a large corporate setting working with a small group of institutional clients, seeing that the Top 25 algos have been cloned over 13,000 times, an average of over 500 clones per strategy is… well it’s pretty damn cool.