Adrian ReidFounder of Enlightened Stock Trading

Which Outliers to REMOVE From Your Trading System Backtest (And Which to Leave In)

15 Sept 2025 · 1 min

Watch on YouTube

0:31

Short excerpt from a YouTube video by Adrian Reid discussing how to handle outliers in trading-system backtests. The speaker suggests removing extreme winning outliers (examples: 300%–1200%) to preserve robustness for typical positions, while keeping extreme losing outliers in backtests to stay pessimistic and reflect real-time trading risk.

Questions this video answers

500%, 400%, 300%, 1000–1200%.

99.9% of positions (with extreme winners occurring ~0.01% of the time).

Chapters

If you have winners that are like 500%, 400%, 300%, then you have like 1000 1200% winner. I might take those out because I want my system to be robust in the 99.9% of positions that it takes. And if it finds one of Those, you know, 0.01% of the time winners, that's fantastic. If you do it long enough, the odds are that it will. And then for outliers, in terms of losers, I always leave those in because it's better to be pessimistic than optimistic. And just like winner outliers can happen. Losers, outliers can definitely happen. Real time trading as well.