Back to Posts
Listen to Thread

is there a python library that allows for the optimization of parameters used within a strategy? I want to parameter step over several years in rolling 30 day periods, then average the results for each parameter and optimize for max return min risk... Otherwise as a brute force method, can I simply loop through running the backtest several times at different parameters and with a few if statements record the optimal parameters at each time interval?

Hi Brandon,

There are plenty of python libraries for general, batch optimization. However, optimizing a trading strategy in an online way isn't quite that simple for several (mostly implementational) reasons. I wrote a blog post on this here which you might be interested in.

As to your second proposal: Yes, you can just optimize by trying out one parameter value and run a whole backtest. I will actually give a talk on how to do this at PyData with zipline (Quantopian's open-source backtester). Unfortunately we haven't gotten around to making this available in Quantopian but it is something we would like to do.

Thomas

Log in to reply to this thread.
Not a member? Sign up!