Independent backtest lab
Test beforeyou trust.
We take the trading strategies everyone shares, code every reasonable version and run them on years of real market data with real fees. Then we publish every number, including the ones that don't sell.
66% of versions look profitable after fees. Compared with simply holding the market and re-tested on data they never saw: 0 proven edges.
The best-backtest trap
Try enough settings and one will look brilliant.
versions of one strategy
We tested 1,249 versions of Moving-average trend following. The typical one makes +0.18R per trade, the best one +2.74R, as much by luck as by skill. That best one is the result that gets shared. We show you all of them.
Moving-average trend following →Average result per trade after fees (R), for the 1,148 versions with enough trades to judge. One bar = a group of versions.
Method
How we test, in 5 steps
The same five checks for every strategy. A real edge has to pass all of them.
Code the rule exactly
Every reasonable reading of the rule becomes a version: stops, holding times, timeframes.
Real costs
Fees on both sides, slippage and funding, on every trade.
Two halves
A rising market and a flatter one: a real edge works in both.
Against the market
We subtract what simply holding the market would have made over the same days.
Data it never saw
The best version is run again on older data (2020–2022) and, when it deserves it, on a held-back year.
The tests
Popular strategies. Same honest test.
Each test comes with the method, every setting tried, the verdict and the full trade list.
Fair value gap (FVG)
No edge after costsMoving-average trend following
No edge after costsNew York open · 12 EMA on Nasdaq
Not provenRSI pullback in a trend
Too weak to tradeIchimoku cloud
No edge after costsTime-series momentum
Too weak to tradeChannel breakout (Donchian / turtle)
Not proven52-week-high anchoring
LessonThe market
A lesson from the lab's data
LessonFees
A lesson from the lab's data
LessonTight stops
A lesson from the lab's data
LessonDoubling down
A lesson from the lab's data
LessonThe best backtest
A lesson from the lab's data
Follow the tests
Test before you trust.
Short videos: one popular strategy, one honest test, the numbers behind it. In English, French and Spanish.