Glossary

Risk of ruin

Last reviewed: 15 September 2026·Tradelyze

Risk of ruin is the probability that a trading strategy's losses reach a loss limit you cannot recover from, such as a prop firm's maximum drawdown or the point where you would quit. It grows with risk per trade: at a 55% win rate with equal-sized wins and losses, risking 1% against a 10% limit, the classic formula gives 13.4%.

New to this? Start with win rate and expectancy.

What is risk of ruin in trading?

Risk of ruin in trading is the probability that losses take an account down to a level where trading has to stop. The idea comes from the gambler's ruin problem in probability theory, where ruin meant losing the whole stake. Traders use the term more narrowly. Wikipedia's Risk of ruin article says financial traders sometimes use it for losses that take an account below its minimum requirements. The article credits that usage to a 2009 chapter by Gregoriou and Rouah on evaluating commodity trading advisors.

Risk of ruin is about the path, not the average. Two strategies with the same average profit per trade can have very different risk of ruin. Ruin depends on how long losing streaks run and how much each loss costs. A losing streak that arrives before the winners can end an account that would have been profitable a year later.

That is why risk of ruin matters so much in a prop firm challenge. A prop firm funds traders who pass its evaluation, and it closes the evaluation account when losses reach its limit. No later winning trade can undo that.

How do you calculate risk of ruin with the classic formula?

The classic risk of ruin formula covers the simplest case. Every win and every loss is the same size, the win rate never changes, and you risk the same dollar amount on every trade. Start by counting how many losing trades your loss limit can absorb. With a 10% loss limit and 1% risked per trade, that cushion is 10 losses.

Write p for the win rate and q for the loss rate, so q = 1 − p. When p is above q, the chance of ever losing the whole cushion is q ÷ p multiplied by itself once for each unit of cushion:

This is the gambler's ruin result given in Wikipedia's Gambler's ruin article, with the market playing an opponent whose money never runs out. The same article states that a persistent gambler with finite wealth, playing a fair game against an opponent with infinite wealth, eventually goes broke. In trading terms: at a win rate of 50% or less with equal-sized wins and losses, the formula gives 100%.

Worked example. A strategy wins 55% of its trades, so q ÷ p = 0.45 ÷ 0.55 = 0.818. Risking 1% against a 10% loss limit, risk of ruin is 0.818 to the power 10, or 13.4%. Risking 2% halves the cushion to 5 losses and raises risk of ruin to 36.7%. Risking 0.5% doubles the cushion to 20 losses and cuts it to 1.8%.

What does the risk of ruin formula assume?

The classic risk of ruin formula rests on six assumptions, and real trading breaks most of them:

• Equal-sized wins and losses. The short formula only works at a 1:1 payoff ratio. A 2:1 strategy needs a numerical solution, used for the 2:1 rows in the risk per trade table on this page.
• A known, fixed win rate. A backtest's win rate is only an estimate. A standard error is the typical gap between a number measured on a limited sample of trades and the true value behind it. Over 100 trades, a measured 55% win rate has a standard error of about 5 percentage points. That is the square root of 0.55 × 0.45 ÷ 100. So a true win rate near 50% is quite possible.
• Independent trades. Real losses tend to cluster, and clustered losses reach a loss limit sooner.
• Fixed dollar risk. Sizing as a percentage of current equity makes each loss smaller as the account falls.
• No limit on the number of trades. The formula gives the chance of ruin over an endless run of trades. Over a shorter run the chance is lower. For the 55% example at 1% risk, an exact calculation for this page gives 9.0% within 100 trades and 11.9% within 200, against 13.4% ever.
• A loss line that stays put. A trailing loss limit rises with the account's peak, so with the same allowance a strategy breaks a trailing limit at least as often as a static one.

Why does ruin mean hitting a drawdown limit rather than losing everything?

Ruin means hitting a drawdown limit because an account almost never goes to exactly zero, and trading stops long before it would. A drawdown is a fall from a previous high in account value. Edward Thorp points out, in his 2006 chapter on the Kelly criterion, that a trader who bets a fixed fraction of current capital can never reach zero. So ruin in the textbook sense cannot happen, and it has to be redefined. The arithmetic shows why. Risking 1% of current equity, 10 straight losses leave 90.4% of the account. It takes 11 losses in a row to fall more than 10%.

In practice the ruin line is wherever the account actually ends. For a personal account that might be a margin requirement, or the point at which you would abandon the strategy. For a prop firm account it is the firm's maximum loss or total drawdown limit. Many firms also add a daily loss limit that can end the account. The rules differ by firm and by drawdown type, static or trailing, as explained in prop firm rules and backtest metrics.

Measure risk in units of the ruin line, not of the full balance. Risking $500 on a $100,000 account looks like 0.5%, but against a $5,000 loss limit it is a tenth of everything you are allowed to lose. Losses also compound against you: a 10% drawdown needs an 11.1% gain just to get back to the previous high.

How does Monte Carlo simulation estimate ruin probability?

Monte Carlo simulation estimates ruin probability by counting instead of solving a formula. It builds many simulated trade sequences out of a backtest's real trades and measures the maximum drawdown of each one. It then reports the share that went past the ruin line:

Counting has two advantages over the classic formula. It uses the strategy's real trade results, so uneven wins and losses, and the occasional very large loss, are included as they happened. And a block bootstrap, which redraws runs of consecutive trades instead of single trades, keeps losing streaks together. Monte Carlo simulation explains the different versions of the test.

A Monte Carlo ruin probability is an estimate, so it never lands exactly on the true chance. In a constructed simulation for this page, 1,000 runs of 500 trades each used a 55% win rate, 1:1 payoff and 1% risk. Of those runs, 132 went past a 10% line, a 13.2% estimate, against an exact 13.3% chance within 500 trades. The standard error is the typical gap between an estimate like this and the true chance. For a share of runs, it is the square root of p × (1 − p) ÷ number of runs, where p is the estimated share. That comes to about 1.1 percentage points at 13% and 1,000 runs.

Monte Carlo simulation cannot add information the trade list does not contain. If the backtest's trades came from one kind of market, or from settings picked out of many tries, every simulated run inherits that.

How do risk per trade, win rate and payoff ratio change risk of ruin?

Risk per trade changes risk of ruin more than anything else, and win rate and payoff ratio decide how steeply it changes. The payoff ratio is the average win divided by the average loss, so a 2:1 strategy wins twice what it loses. Expectancy is the average result per trade, written here in R, where 1R is the amount risked on one trade. In R, expectancy = win rate × payoff ratio − loss rate. Win rate and expectancy explains both terms.

This page uses no JavaScript, so instead of a risk of ruin calculator the table gives worked values against a 10% loss line. Each value is the chance of ever losing 10% of the starting balance. The model uses a fixed dollar risk per trade and independent trades, with every loss exactly 1R and every win exactly the payoff ratio.

Three readings stand out. First, risk per trade dominates: at a 55% win rate and 1:1 payoff, going from 1% to 2% risk takes risk of ruin from 13.4% to 36.7%. Second, expectancy alone does not decide it. A 60% strategy at 1:1 and a 40% strategy at 2:1 both earn +0.20R per trade. Yet at 1% risk their risk of ruin is 1.7% and 14.2%, because a lower win rate brings longer losing streaks. Third, a thin edge needs very small size: at 35% and 2:1, even 0.25% risk leaves a 13.7% risk of ruin.

What changes when a prop firm challenge has a profit target?

A profit target ends a prop firm challenge when it is reached, so the question becomes whether the account hits the loss line before the target. The gambler's ruin result with two finish lines answers that, and the chance is always lower than the chance of ever hitting the loss line. With a 10% loss line and a 10% profit target, an exact calculation for this page gives these figures, a constructed illustration and not measured data:

• 55% win rate, 1:1 payoff, 1% risk: 11.9%, against 13.4% with no target.
• 55% win rate, 1:1 payoff, 2% risk: 26.8%, against 36.7%.
• 40% win rate, 2:1 payoff, 1% risk: 12.6%, against 14.2%.
• 50% win rate, 1:1 payoff, any size: 50%, because with no edge two lines the same distance away are equally likely to be hit first.

Smaller size has a cost. At 0.5% risk the 55% strategy needs twice as many net winning trades to reach the same target, so the challenge takes roughly twice as many trades. These figures also ignore daily loss limits, minimum trading days and consistency rules. Position sizing for prop firm challenges turns a firm's limits into a trade size.

What risk of ruin is acceptable?

Trading calculator and education sites commonly quote a risk of ruin below 5% as the target, and below 1% as the level professionals aim for. No primary source sets a safe level of risk of ruin for trading. The pages that repeat 5% and 1% give no study or derivation behind either figure. Treat any single figure quoted that way as one writer's preference, not a finding. What exists instead are competing ways of deciding, and they pull in opposite directions.

Thorp writes that "most cautious gamblers or investors who use Kelly find the frequency of substantial bankroll reduction to be uncomfortably large". Such users, he adds, tend to bet less than the full Kelly fraction. Full Kelly also clashes with a hard loss limit. Wikipedia's Kelly criterion article gives the formula for the Kelly fraction. For a strategy that wins 55% of the time at 1:1, it is 0.55 − 0.45 ÷ 1, or 10% of the account per trade. One loss at that size would end an account with a 10% loss limit.

A practical way to decide is to price the ruin. A failed challenge costs a known, capped fee; a blown personal account that took years to build does not come back. The more a breach would cost you, the lower the risk of ruin you should accept, and the lever you control directly is size.

How do you read Tradelyze's Ruin Probability?

Tradelyze's Ruin Probability is the share of Monte Carlo runs whose drawdown went deeper than the total drawdown limit of the rule set you selected. It sits in the Monte Carlo rows of the robustness card, beside MC Max DD Real→P95. By default Tradelyze builds 1,000 runs by redrawing your trades with a stationary block bootstrap. That method strings together random runs of consecutive trades, so losing streaks stay intact. The trades are those of the chosen settings over the history the optimizer searched, so the figure is in-sample.

Four details decide how to read Tradelyze's Ruin Probability:

• It is counted per rule set. Each selected rule set's figure is re-counted from the same simulated drawdowns at that rule set's own limit. A tighter limit therefore shows a higher figure for the same strategy. Do not compare Ruin Probability between firms.
• It is measured from the peak. Each run's drawdown is measured from its highest equity, even when the rule set's own total drawdown rule is static and measured from the starting balance. On a static rule set that makes Ruin Probability a stricter test than the matching row in Rule Results.
• It leaves out daily limits. A redrawn run has no calendar, so daily loss limits are checked separately on the real trade order in each firm's Rule Results table.
• It can miss a dip inside a bar. Where the price data allows, Tradelyze uses each trade's worst point while it was open, taken from bar highs and lows. A dip that happened and recovered inside one bar is not seen.

Ruin Probability also moves the robustness score. The Monte Carlo check is worth 25 of the 100 points, scored as 25 × (1 − Ruin Probability ÷ 50). So 10% earns 20 of those points, 20% earns 15 and 50% or more earns none. Sometimes the Monte Carlo check runs but no Ruin Probability can be counted for a rule set, for example because it has no total drawdown limit. Tradelyze then leaves the score, grade and verdict blank for that rule set rather than grading at another limit.

Worked example. In a constructed case with a 10% total drawdown limit selected, suppose 132 of 1,000 runs fell more than 10% from a peak. The row reads 13.2% in green, and the Monte Carlo check earns 25 × (1 − 13.2 ÷ 50) = 18.4 of its 25 points.

How do you lower risk of ruin before a prop firm challenge?

Lower risk of ruin by trading smaller, because size is the one input you control directly. In the risk per trade table on this page, halving risk from 1% to 0.5% cuts a 55%, 1:1 strategy's risk of ruin from 13.4% to 1.8%. A practical order of work:

1. Find the real ruin line: the firm's total drawdown limit, whether it trails, and its daily loss limit. Trailing drawdown explains why a trailing line is harder to survive.
2. Size from the Monte Carlo worst case, not from the backtest's single drawdown. Sizing off the Monte Carlo drawdown explains why.
3. Turn the limit into a trade size with the arithmetic in position sizing for prop firm challenges.
4. Set that size in the strategy itself. In a TradingView Pine Script strategy, the default order size is set in the strategy() declaration with default_qty_type and default_qty_value; TradingView strategy properties covers the settings.
5. Run the backtest and the Monte Carlo test again, and compare risk of ruin at the new size.

Do not try to lower risk of ruin by re-tuning settings until the backtest win rate rises. A win rate raised by tuning on the same history tends to fall back in live trading. The risk of ruin formula would then be fed a number that was never real. Overfitting and sample size explains why.

What can risk of ruin not tell you?

Risk of ruin cannot tell you whether the trades behind it are real. Every risk of ruin figure, from the classic formula or from Monte Carlo simulation, starts from a backtest's win rate, payoff ratio or trade list. Settings picked as the best of many tries flatter those inputs. The risk of ruin that comes out is flattered with them: overfit trades in, optimistic ruin out. Tradelyze's Ruin Probability is built from in-sample trades, so treat it as optimistic rather than conservative.

A small error in the inputs makes a large error in risk of ruin. Take 1% risk against a 10% line, with equal-sized wins and losses. The classic formula gives 13.4% for a 55% win rate, 44.9% for a 52% win rate and 100% for 50%. Yet over 100 trades, a measured 55% and a true 52% are easily confused. At that sample size the standard error, the typical gap between a measured win rate and the true one, is about 5 percentage points. The two differ by only 3.

Risk of ruin also leaves out:

• Losses bigger than planned, such as a price gap through a stop.
• Markets that behave differently from the backtest period.
• Daily loss limits, time limits and other rules that can end an account early.
• Whether the strategy makes money at all. A low risk of ruin says the losses are small relative to the limit, not that the edge is real. The other checks behind the robustness score, and maximum drawdown, cover different parts of that question.

Frequently asked questions about risk of ruin

Sources

• Wikipedia, Gambler's ruin, retrieved 15 September 2026, en.wikipedia.org/wiki/Gambler's_ruin — the unfair coin flipping result for the probability of each player ending penniless, from which the classic formula follows as the opponent's bankroll grows without limit, and the statement that a persistent gambler with finite wealth, playing a fair game against an opponent with infinite wealth, eventually goes broke. The article refers to Feller (1970), An Introduction to Probability Theory and Its Applications, 3rd edition, for the method; Feller was not consulted directly for this page.
• Wikipedia, Risk of ruin, retrieved 15 September 2026, en.wikipedia.org/wiki/Risk_of_ruin — the narrow trading usage of the term for losses that reduce an account below minimum requirements, which the article attributes to Greg N. Gregoriou and Fabrice Douglas Rouah, A Risk of Ruin Approach for Evaluating Commodity Trading Advisors, chapter 22 in Operational Risk Toward Basel III, John Wiley & Sons, 2009, page 453. That chapter was not consulted directly for this page.
• Edward O. Thorp, The Kelly Criterion in Blackjack, Sports Betting, and the Stock Market, chapter 9 in S. A. Zenios and W. Ziemba (editors), Handbook of Asset and Liability Management, Volume 1, Elsevier, 2006, pages 385–428, DOI 10.1016/S1872-0978(06)01009-X (confirmed on 15 September 2026 to resolve to the Elsevier record; page range from Crossref); first presented at the 10th International Conference on Gambling and Risk Taking, Montreal, June 1997 — that fixed-fraction betting cannot reach zero, so ruin must be redefined; that minimum bets minimize both the probability of ruin and expected gain; section 3.2, where the probability of ever being reduced to a fraction x of the initial bankroll equals x at the Kelly fraction in the continuous approximation; and the quoted observation about cautious Kelly users.
• J. L. Kelly, A New Interpretation of Information Rate, Bell System Technical Journal 35(4), 1956, pages 917–926, DOI 10.1002/j.1538-7305.1956.tb03809.x (confirmed on 15 September 2026 to resolve to the IEEE Xplore record). The betting-fraction formula f = p/l − q/g used on this page is as given in Wikipedia, Kelly criterion, retrieved 15 September 2026.
• Trading risk of ruin calculator and education pages, retrieved 15 September 2026. Four calculator pages were read in full. All four gave a risk of ruin below 5% as the target or acceptable level, and described below 1% as the professional, institutional, “excellent” or “safe” level. None gave a study or derivation for either threshold. They are not named or linked here because they are not primary sources.
• Dimitris N. Politis and Joseph P. Romano, The Stationary Bootstrap, Journal of the American Statistical Association 89(428), 1994, pages 1303–1313 — the block bootstrap that Tradelyze's Monte Carlo check uses.
• Tradelyze constructed calculations for this page, 15 September 2026. Constructed illustration, not measured data. Rows with a 1:1 payoff use the closed form (q ÷ p) ^ cushion. Rows with a 2:1 payoff use the probability r of ever falling by one unit of risk, solved from r = q + p × r³ and raised to the cushion, cross-checked by a step-by-step probability calculation. The within-100-trades, within-200-trades and profit-target figures use the same step-by-step calculation, and the Monte Carlo illustration is one seeded simulation of 1,000 runs of 500 trades.
• Tradelyze Monte Carlo and ruin implementation, reviewed 15 September 2026: 1,000 simulations by default; a stationary block bootstrap on the chosen settings' trades over the optimization data; each run's maximum drawdown measured peak to trough as a percentage of peak equity, at bar resolution where the price data allows; Ruin Probability is the percentage of runs whose maximum drawdown exceeds the selected rule set's total drawdown limit, re-counted for each rule set; blank when no usable limit exists, when the check refused to run, or when no simulated drawdowns were recorded; the Monte Carlo check fails at 20% or more; Monte Carlo points are 25 × (1 − ruin ÷ 50), floored at zero; score, grade and verdict are blank when the check ran but no ruin figure can be counted for the rule set; daily loss limits are checked separately on the real trade order.