I'd like to share experiences here how to validate backtest results on out of sample data which could then go for real trading. I'd be interested in what KPIs you look at and for what threshold it needs to fulfill in order to select a system for real trading. Please try to reply if you have successfully validated your system and successfully traded it over years. I am not really interested in theories but mostly the ideas that have been verified in practice. Please also indicate what instrumentum and what time frame you used as I think a system on FX 5-minute database has different validation method than stocks on EOD.
For example for stocks, EOD data I found the following KPIs useful for trading system validation.
- Profit Factor > 2.7
- Ttest > 3 (this is not in the default report)
- smooth equity curve and short draw down periods (this is difficult to specify exactly)
- MC min equity in Report/Monte Carlo sheet should also be climbing up
How do you validate your OOS backtest results?