Strategy Stress Test
Try to prove your own strategy wrong before the market does. Enter a record, or your whole trade history, and see whether the edge is statistically real after fees, how many trades would settle it, and which prices leak.
Quick Answer
A record is an edge only when it clears break-even after fees by more than luck explains. Worked example: Not proven yet: 59% wins against a 54% break-even over 41 trades. About 287 trades would settle it.
Every version you tried and dropped is another roll of the dice. The bar rises with each one (a Bonferroni correction), so a strategy picked from 20 tries has to clear a much higher bar than a strategy you wrote down first.
Not proven yet
Not proven yet: 59% wins against a 54% break-even over 41 trades. About 287 trades would settle it.
Win rate
58.5%
interval 45.7–70.3%
Break-even
53.7%
price + fee per contract
Trades needed
287
at 1 version tried
How the verdict is read
Net per dollar staked, not pooled win rate. A win rate means something only against the price you paid for it. Mix 90¢ favorites with 5¢ longshots and the pooled win rate can sit well above the pooled break-even on a book that is down after fees. So the headline verdict reads your net result per dollar staked, with an interval, and the win-rate-versus-break-even comparison appears only inside each price band.
Positions, not contracts. Ten contracts bought on one market are one decision, so every interval counts positions. A Kalshi export lists closed lots; the tool groups them by market and side first.
Fees in. Kalshi fees come from the published schedule (7% × price × (1 − price) per contract, taker; see the Kalshi Fee Calculator). For other venues, include your fees in the file; without them the tool assumes zero and says so.
Both halves. With timestamps, the history is split at its midpoint. An edge that lives in one half and not the other is a streak, not a strategy. Next step for an edge that holds up: size it with the Kelly calculator.
Frequently Asked Questions
How many trades do I need to know if my prediction market strategy works?
It depends on how far your win rate sits above break-even. The test is the number of trades at which the gap stops being explainable by luck: z² × p(1 − p) ÷ (p − break-even)², with z = 1.645 for one strategy at 95% one-sided confidence. At 24 wins in 41 Kalshi trades bought around 52¢, that works out to about 287 trades. A small edge needs hundreds of trades; at or below break-even, no number of trades proves one.
What win rate do I need to break even on Kalshi?
Your break-even win rate is the price you pay plus the fee, in cents, as a percentage. Kalshi's taker fee is 7% × price × (1 − price) per contract, charged win or lose, so a 52¢ contract costs about 53.75¢ all-in and needs a 53.7% win rate to break even. Compare win rate to break-even inside one price band at a time: pooled across very different prices, the comparison can show an edge on an account that is losing money.
Why do cheap Kalshi contracts lose money?
In a study of 300,000+ Kalshi contract prices (2021–Apr 2025), buyers of contracts under 10¢ lost over 60% of their money on average. A cheap contract pays a lot when it wins, so a few wins make a long run of small losses look fine, and the fee is a bigger share of a small price. That is why this tool reports results by price band and asks for more trades before it calls a sub-10¢ edge real.
What does 'strategies tried' change?
Every version you test and drop is another chance to get lucky. The tool raises the bar for each one with a Bonferroni correction: one strategy needs z = 1.645, ten need z = 2.576. The same 24-for-41 record at 52¢ reads "not proven yet" with one try and needs 703 trades to settle at ten.