Quantopian's community platform is shutting down. Please read this post for more information and download your code.
Back to Community
Compare many trading strategies, each with many variants

I'm developing a custom Python quant program and have 100 Trading Strategies (e.g. Keltner Channesl, Bollinger Bands, MACD, etc). I also have 10 parameters that I cycle through for each strategy (e.g. Fast_Periods, Slow_Periods, Trailing Stop %, etc). When I run the simulation, it generates about 500,000 combinations of Trading Strategies and Parameters Sets. I run the simulation over all S&P500 companies over 20 years of daily data. So far, so good - it works. My issue is analyzing all that output. I port is to pandas and Excel.

I refer to a Trading Strategies averaged over all Parameter Sets as a "Parent Trading Strategy" (PTS).
I refer to a Trading Strategy with a specific set of parameter values as a "Trading Strategy Instance" (TSI).

I find the top 10 PTS are different from the top 10 TSI. Are there any best practices for comparing Trading Strategies overall (PTS), vs specifically parameterized Trading Strategies (TSI).

I can compare risk adjusted returns of each strategy or each PTS and of each TSI, and even of each set of parameter values. But I'm unclear how to aggregate them in to an overall conclusion of which to select for actual trading. Has anyone else faced a similar problem?
Thanks in advance
Phil