Quantopian Challenge

I’m going to take a small detour this month to discuss the Quantopian Challenge.  Quantopian runs an online investment algorithm contest.  You code your algorithm on their website (using the Python language and their libraries), and enter it into a monthly contest where you compete against other algorithm writers for the ‘best’ algorithm.

‘Best’ is subjective, of course.  Return on investment is an obvious criterion, but it isn’t the only one.  They’re also looking for a low correlation compared to the SP500 index (low Beta), use of hedging (having both long and short positions), and other criteria.  In fact, they don’t tell you exactly what all of the criteria are, or how they’re weighted to reach a final score.  But it is this final score that they use to declare the ‘best’ algorithm.

No actual money is traded up to this point, so you don’t have to worry about having capital (or losing it).  The interesting part is that Quantopian also runs a hedge fund, which does trade real money.  If they like your algorithm’s results, they may give it a capital allocation from their hedge fund, and give you a percentage of any profit it generates while they’re using it (but not any losses, so again, nothing to worry about there).  So, for example, they might trade $2 million of their fund using your algorithm.

It’s a clever idea.  They believe they can get a fairly large number of sophisticated, but non-professional, investors to write algorithms with the attributes they’re looking for, and in return they pay said investors a percentage of profits they generate.

Last Fall I spent some time coding one of my existing algorithms into their system and entered it into a contest.  Contests start once a month, and run for six months, so the first contest just ended.  My entry came in 11th place.  At the end there were about 320 algorithms in the contest.  There were more at the beginning, about 500 if I remember correctly.  The contest that just started in March has 1500 entries, so I guess it’s getting more popular.

(NOTE: When I looked at the just-ended contest earlier today it said I came in 12th, now it says 11th.  I have no idea why.  Maybe in a few weeks I’ll break into the top 3.)

The same algorithm automatically gets entered every month now.  I’ll work on tweaking it a bit, and see if I can top my existing algorithm.  You  can enter up to 3 algorithms at a time, so no need to shut down my current algorithm.  And I’ll update you on how I’m doing.

If you decide to enter the contest, please drop me an email and let me know.  I won’t ask for any details (or provide any of mine), but I do like to talk to people about investment strategies.