Risk of ruin & statistical significance: how many trades to trust an edge
A great month doesn't prove an edge; a bad week doesn't kill one. The maths of sample size and ruin probability tells you which of your results is signal and which is noise.
Two questions every trader should be able to answer about their strategy: how many trades before I know it works? and what's the chance it wipes me out before I find out?
Sample size: when noise stops dominating
A 30-trade sample tells you almost nothing — the confidence interval around the win rate is huge. Reasonable statistical confidence in a real edge typically needs hundreds of trades. Below ~100, results are dominated by variance; above ~300, the edge (if any) starts to surface clearly. This is why a single great evaluation isn't proof of skill.
Risk of ruin — the three levers
The probability of blowing the account before the edge plays out depends on three things, not one:
- Edge — win rate combined with payoff ratio (avg win / avg loss).
- Risk per trade — what fraction of capital each trade puts at stake.
- Bankroll / drawdown buffer — how many full losses the account can absorb.
The bigger your edge, the lower your ruin probability. But ruin is exponentially sensitive to risk per trade: doubling risk doesn't double ruin probability, it can multiply it many times over.
Why small edge + low risk beats big edge bet hard
A modest edge sized conservatively can have near-zero ruin probability. A big edge sized aggressively can still have a meaningful chance of blowing up before the edge converges. On a prop account, the drawdown floor is your bankroll — small, and unforgiving. That's why drawdown-based ratios like Calmar matter more than headline return.
How to actually estimate it
Closed-form ruin formulas exist but they assume fixed win% and payoff — rarely true. The honest method is Monte Carlo simulation: resample your trade outcomes into thousands of alternate sequences and count the share that hit ruin under your account's real rules. That number — the modeled blow rate — is the single most important metric for a funded trader. See how to know if a backtest is overfit and the methodology page.
Puravida Edge reports annualized blow rate from a 1,500-path Monte Carlo per portfolio — not just a return. Check it for any setup in the Pass Estimator, and size accordingly with the Position Size Calculator.
FAQ
How many trades do I need to know if my strategy works?
A 30-trade sample is noise; reasonable confidence usually needs hundreds. Below ~100 trades, variance dominates; above ~300, a real edge starts to surface clearly.
What is risk of ruin in trading?
The probability of blowing up the account before the edge plays out. It depends on edge size (win rate & payoff), risk per trade, and bankroll — and grows exponentially with risk per trade.
How do I calculate risk of ruin honestly?
Use Monte Carlo simulation: resample trade outcomes into thousands of sequences and count how many hit ruin under real account rules. The resulting modeled blow rate is the most honest measure.
Not financial advice. Performance figures referenced are hypothetical, modeled outputs (1,500-path Monte Carlo on a 12-month sample). Past performance does not guarantee future results. Tool names are referenced for education; verify current features and prop-firm rules directly.