What "risk of ruin" means here
Risk of ruin is the probability that a trading account falls to a level you cannot accept. This simulator measures a fixed floor below your starting balance (for example 20% down) within a fixed number of trades. A path counts as ruined the first time it touches the floor, even if it would have recovered later. It does not measure the account reaching zero: when you risk a fixed fraction of current equity, the balance shrinks after each loss but never reaches exactly zero, which is why a practical floor is the useful question.
How the simulation works
- Each path starts at equity 1 (100% of the starting balance).
- For each trade a random number decides win (probability p) or loss.
- A win multiplies equity by 1 + f × R; a loss multiplies it by 1 − f, where f is the fraction of current equity at risk and R the average win in multiples of that risk.
- If equity is at or below 1 − floor, the path is marked as ruined and stops.
- The estimate is the share of paths that were ruined, with a 95% Wilson interval for simulation error.
The same random draws are used for your risk level and for half of it, so the difference between the two columns comes from position size, not from luck in the sample. The generator is a seeded xorshift32, so any result can be reproduced exactly.
Example
With the defaults (55% win rate, 1.5R average win, 1% risk, 100 trades, 20% floor, 5,000 paths, seed 42), none of the 5,000 paths touches the floor. That does not mean the risk is zero: with 0 of 5,000 the 95% upper bound is about 0.08%, and the result depends entirely on the assumed win rate and payoff being right. A very simple check of the logic: with a 50% win rate, 1R wins, 50% risk and a 50% floor over two trades, exactly half of all sequences start with a loss and hit the floor at once, so the true risk is 50%; the simulator returns 2,452 of 5,000 paths (49.0%), within sampling error.
Assumptions and limits
- Trades are independent and the win rate and payoff do not change. Real strategies have losing streaks clustered by market regime.
- Every win is exactly R and every loss exactly the amount risked. Real outcomes vary, gaps can make losses larger than planned, and fat tails are not modelled.
- Costs must already be included in R and the risk per trade.
- Your win rate is itself an estimate. A 55% rate measured over 40 trades is consistent with a much lower true rate; see the uncertainty range in the breakeven win rate calculator.
- This is about trading drawdowns, not retirement withdrawals, which are a different model.
FAQ
What is a good risk of ruin?
There is no universal number. It depends on how much drawdown you can live with and how confident you are in your edge. Use the simulator to see how the probability changes with risk per trade, and remember it is only as good as the inputs.
Why does halving the risk per trade cut the risk of ruin so much?
At 1% risk, 23 losses in a row take the account below a 20% floor (0.9923 ≈ 0.794); at 0.5% risk it takes 45 (0.99545 ≈ 0.798). With wins mixed in, the same logic holds: the account needs a much longer bad run. Long losing runs are far rarer than short ones, so the probability falls faster than the position size.
Zero paths hit the floor. Is my risk zero?
No. A finite simulation can miss rare events, and the model leaves out regime changes, correlated trades and gap losses. Read the upper bound, not the zero, and expect real-world risk to be higher than this simplified model shows.
How many simulations do I need?
More paths narrow the interval: with 10,000 paths and a true risk near 5%, the 95% interval is roughly ±0.4 points. The cap of 5,000,000 trades simulated in total keeps the page responsive on phones.
Educational simulation, not trading advice or a forecast. Trading with leverage can lose more than you expect; past or simulated results do not guarantee future results.