True AI Trading · Tools

Prop firm evaluation tools

An evaluation is a path-dependent, single-shot event with an absorbing barrier - not a backtest. Replaying your history once tells you almost nothing, because you only ever observed one ordering of your trades. These tools simulate thousands of orderings instead, and show you the one number the industry never publishes: how often a strategy with no edge at all passes.

-
probability of passing, across 10,000 independent orderings of your own trade distribution
Enter your numbers and run the simulation.

Simulated equity paths

30 sample runs. Paths stop the moment a rule is breached - that's the absorbing barrier doing its work.
Passed Failed

What happens across all 10,000 runs

Every run ends exactly one way. Bar length is the share of runs.

Why one replay of your history is misleading

Run your actual trade sequence through the rules once and you get a single yes or no. But you only observed one ordering. Shuffle the same trades - same win rate, same average win, same edge - and a losing streak that happened to land in month three now lands in week one, when your buffer is thinner, and the account is gone.

The drawdown clause usually matters more than the strategy. Switch the type from intraday-trailing to static and watch the number move. Same edge, same trades - different contract. That's why shopping rule sets beats shopping brands, and it's the entire point of the comparison tab.

The no-edge baseline is the number nobody shows you. Feed in a strategy with literally zero expectancy - same trade sizes, breakeven win rate - and it still clears a typical evaluation roughly one time in five. Enough people buy enough attempts that a steady supply of funded accounts belongs to traders with no edge whatsoever. If your own number isn't well clear of that line, passing tells you nothing about your strategy.

And note what these tools do not tell you: whether your edge is real. Every number assumes your win rate and average win survive out of sample. If they came from a backtest you iterated on, they probably won't.