Risk of Ruin Calculator

How the risk of ruin calculator works

Risk of ruin asks one question: starting from today, what is the chance that bad luck alone takes the account down to a level I cannot accept? Here the level is the ruin floor, a percent of the starting balance. A floor of 50 means the account has halved. You pick it. Zero would mean total loss, which a percent-risk model never reaches exactly, so a floor above zero is more useful.

The calculator does not use a closed formula. It plays the strategy forward 5,000 times with a seeded random number generator (mulberry32, seed 123456789), so the answer is the same on every run and in every browser. On each path every trade is a win with the chance you set and a loss otherwise. The balance is updated trade by trade.

A worked example

Take a made-up strategy with a 40 percent win rate and a payoff ratio of 1.6, risking 2 percent of the balance per trade for 300 trades, with a floor of 50. A win adds 2 x 1.6 = 3.2 percent and a loss takes off 2 percent. The average result per trade is 0.4 x 1.6 - 0.6 = 0.04 times the amount risked, so the odds are positive. Even so, in the 5,000 simulated paths 6.2 percent fell to half the starting balance at some point. The median path ended at 115.4 percent of the start, the 5th percentile path at 56.0 percent, and the longest run of consecutive losses in any path was 26.

Now change only the size. At 1 percent risk the chance of touching the floor is 0.0 percent, at 4 percent it is 37.6, at 6 percent 61.4, and at 10 percent 86.0. The odds did not change. Only the share of the balance risked did.

A fixed check you can do by hand: with a 0 percent win rate, 35 losses at 2 percent leave 0.98 ^ 35 = 49.3 percent of the start, just under the floor, so ruin is certain. With 34 losses it is 0.98 ^ 34 = 50.3 percent, just above it.

What the model assumes, and where it breaks

• Trades are independent. Each trade has the same chance of winning whatever came before. Real markets have trends and quiet spells, so losses cluster more than coin flips do.
• The odds are fixed. The win rate and payoff do not change over the run. A strategy that stops working, or was overfitted to the past, does not stay at its backtest odds.
• Size is a fixed percent of the current balance. The bet shrinks after a loss and grows after a win.
• Costs are only in if you put them in. Enter a win rate and a payoff ratio that are net of spread, commission and slippage.
• Losses are the size you entered. Gaps and stop failures can make a real loss larger than planned.

Why martingale and grid win rates do not fit

A martingale bot often reports a win rate above 90 percent. Typing that into this tool gives a misleading picture, because the model breaks on both assumptions. The bot's trades are not independent: its next size depends on the last result, and the one big loss is the end of a ladder the earlier wins were the first steps of. And the size is not a fixed share of the balance, since it doubles after each loss. Run the ladder through the martingale calculator instead and look at the margin. For the odds of long losing runs, the guide risk of ruin has the arithmetic.

Reading the other tiles

The median final balance is the middle path. Half the paths ended below it. The 5th percentile shows the bad end: one path in twenty did no better. The longest losing run is the longest in any of the 5,000 paths, so it is larger than what you would see in one run. A run that long is not unusual across many trades, which is why a losing streak that "should never happen" appears in practice. To see the full range of outcomes, use the Monte Carlo simulator. To check that the odds are positive before anything else, use the expected value calculator.