The Research
We publish the strategies that failed.
Most backtests you read are the one that worked. Every number in this series came from testing real strategies against real market data, including the eighty-one that didn't survive.
I Backtested 83 Trading Strategies. Two Survived. Neither Was Found By Searching.
83 strategies tested on real MNQ and QQQ data, split train/test. 2 survived — 2.4%. The full denominator, not just the winners.
Is My Opening Range Breakout Overfit? Here's The Ten-Minute Test.
A parameter sweep across the ORB window length on real NQ data — the ten-minute test that tells you if you fit a real effect or a coincidence.
The Same Strategy, Four Hours Later, Loses Money
Identical opening-range-breakout rules, moved from the 09:30 open to the 14:00 session. Same data, same risk model. It loses money.
I Found A Strategy That Works. Buy-And-Hold Still Beat It By 5x.
10.6 years of QQQ data, a real surviving swing setup, and the honest comparison against simply buying and holding.
How To Tell If Your Backtest Is Luck
116 t-statistics from 62 designed strategies, plotted against what pure random noise produces. This is the distribution a real backtester needs to beat.