Quantopian's community platform is shutting down. Please read this post for more information and download your code.
Back to Community
Backtesting - Optimization Methods

In regards to backtesting, is it recommended to perform optimization with 1 parameter at a time rather than multi parameters? Especially if i want to identify what those changes have on the overall output. Currently working on an equity EOD strategy with various parameters which could be set to different values within a predefined range.

Thanks for the input!

2 responses

Hey Adam, can you give more of a description of the parameters you are trying to fit and the model you are using? If you are fitting parameters manually by running a backtest and recording a result each time a parameter is tweaked, it may be better to perform optimization one parameter at a time. Quantopian currently does not have an engine for parameter optimization or Monte Carlo simulation, but you may want to build on ideas other users have posted to do write your own parameter fitting code. Here is one example: https://www.quantopian.com/posts/monte-carlo-simulation

Hi Ryan,
Basically want to run some step increments with params such as RSI, ATR, profit take methods, risk mgmt methods, position sizing methods. I was figuring to run one parameter at a time, however seems to be somewhat inefficient, especially since there may be other relationships which would perform better using multi param step testing.

Thanks for the suggestion on the monte carlo sim. Will take a look.