The Lab
Break a strategy yourself.
Four experiments that run in your browser on simulated prices. No market data is used or downloaded. The point of each one is to make a trap visible before it costs you money.
Experiment 01 · Selection bias
The Null Machine
The prices below contain no edge at all. The machine invents random trading rules, keeps the one with the best in-sample result, then shows you what it does on the 30% of data it never saw. Raise the number of rules and watch the in-sample winner get better while its future stays the same.
Searching …
Simulated: 1,500 trades with normally distributed R and mean zero, six random features; rules are 1-3 interval conditions with at least 60 in-sample trades.
Experiment 02 · Win-rate illusion
The exit makes the win rate
Each line is a trade entered at random on a market with no drift, with a stop of 1 R and a target you choose. Small targets win often, big targets win rarely. The average result is always the same: minus your costs.
Experiment 03 · Costs
The cost fraction
Your edge is measured in R, your costs in points. The tighter the stop, the more of every R the round turn eats. Find the stop below which a real edge turns into a guaranteed loss.
1.75 points is roughly one micro Nasdaq round turn: $2.50 commission plus one tick of slippage each way.
Experiment 04 · Prop-firm math
The evaluation simulator
Two thousand accounts trade the same strategy under a typical evaluation: reach +$3,000 before the balance falls $2,000 below its highest end-of-day value. Same edge, very different fates.
This edge:
Simplified: each trade wins the reward or loses 1 R minus costs, independently; the floor trails the highest end-of-day balance by $2,000 and never locks; no daily loss limit.