This article explains risk of ruin as a concept. It is educational and general, a description of a survival constraint and what drives it, not personalised investment advice and not a claim that any strategy, win rate, or bet size is profitable or suitable for any individual. It does not tell anyone how much to risk or how large a position to take. Nothing here is a recommendation to buy or sell any security.
There is a number that sits underneath every trading system and quietly decides whether it ever gets to prove itself: the chance that a run of bad luck drains the account past the point of recovery before the edge can show up. It is called the risk of ruin, and its most unsettling property is that a strategy can have a genuine edge and still be almost certain to blow up, purely because the bets are too big. Here is what risk of ruin is, why position size dominates it, and why surviving is a precondition for compounding rather than a footnote to it.
The short version
Risk of ruin is the probability that a sequence of losses draws an account down to a level from which it cannot recover, before its long-run edge can play out. It rises with a lower win rate and a worse payoff ratio, but it is dominated by position size: bet a large fraction of the account on each trade and even a positive-expectancy strategy can face a high, sometimes near-certain, chance of ruin.
What risk of ruin actually is
Risk of ruin is a probability, not a prediction. It answers a single question: given how often a strategy wins, how much it makes when it wins versus loses, and how much of the account rides on each trade, what are the odds that a losing streak carries the balance down to a floor it cannot climb back from? That floor might be literal (the account reaches zero, or a broker’s minimum) or practical (a drawdown so deep the operator stops, or so deep the remaining capital cannot mathematically recover the loss). Either way, ruin is an absorbing state: once it is reached, the game is over, and no future edge can undo it.
The key move is to separate two things that are easy to conflate. Expectancy is what a strategy earns on average per trade over the long run. Ruin is about the path it takes to get there. Average outcomes are computed over many trades; ruin is decided by the worst stretch along the way. A strategy can have a perfectly good average and still hit the absorbing floor during an ordinary run of losses, at which point the good average becomes irrelevant, because the account is no longer in the game to collect it.
The gambler’s-ruin intuition
The concept has a clean lineage in probability theory: the classical gambler’s ruin problem, studied since the seventeenth century by Pascal, Fermat, and later Huygens and the Bernoullis. A gambler with a finite bankroll makes repeated bets against an opponent with deeper (sometimes effectively infinite) pockets. The mathematics gives a stark result: if the game is even slightly unfavourable, ruin is eventually certain. If the game is even, the gambler with the smaller bankroll is still ruined with high probability against a much larger opponent, simply because the smaller purse runs out first during a normal swing.
The lesson that carries straight into markets is that a finite bankroll facing repeated bets can be wiped out by variance alone, even when the odds are not against it. The market is the deep-pocketed opponent. The account is the finite purse. And the size of each bet, relative to the purse, decides how much variance the account can absorb before it hits the floor. That relationship, between bet size, bankroll, and the probability of touching zero, is the heart of risk of ruin.
What drives it: three inputs, one that dominates
Three quantities set the risk of ruin, and it is worth being precise about how each one pushes it.
- Win rate. The fraction of trades that are winners. Holding everything else equal, a lower win rate means longer expected losing streaks, and longer streaks are more likely to reach the floor. Win rate matters, but on its own it is a poor guide: a low win rate can be perfectly survivable if the winners are large, which is why it cannot be read in isolation.
- Payoff ratio. The size of the average win relative to the average loss (sometimes folded together with win rate into expectancy, the average outcome per trade). A larger payoff ratio cushions a low win rate: rare large winners refill the account between losing runs. A strategy that wins seldom but big and one that wins often but small can share the same expectancy and yet face very different streak risk.
- Position size, the fraction of the account risked per trade. This is the dominant lever, and the one people underestimate. Doubling the fraction risked on each trade does not merely double the risk of ruin; because losses compound, it can raise it disproportionately, turning a survivable strategy into a doomed one without changing the win rate or payoff at all. Size is the input that converts an edge into either a durable compounder or a fuse.
Because the first two are properties of the strategy and the third is a choice made independently of them, the practical takeaway is that ruin is largely something the operator controls. An operator does not get to pick the market’s win rate, but does choose how much of the account is exposed to any single outcome, and that choice, more than any other, decides whether a real edge ever compounds or quietly detonates.
The counter-intuitive core: a good edge can still ruin you
Here is the result that surprises people, and it is the single most important idea in the topic: positive expectancy does not guarantee survival. A strategy can make money on average, trade after trade, and still carry a high probability of ruin if the bets are sized too aggressively. The edge tells you the balance drifts upward over a long enough horizon. It says nothing about whether the account survives the drawdowns along the way to collect that drift.
The mechanism is compounding, working in reverse. Losses do not subtract in a straight line; they multiply down. A position sized at a large fraction of the account, hit by an ordinary losing streak, shrinks the base that every future trade is sized from, so each subsequent loss removes a larger share of what remains and the hole deepens faster than intuition expects. Past a certain bet size, the drawdowns a positive-edge strategy produces in the normal course of business are deep enough that the absorbing floor is reached before the edge can pull the balance back. The edge is real; the sizing kills it first. This is the precise sense in which, for a leveraged or over-sized bettor, being right on average is not enough. One clarification keeps the picture honest: in an unleveraged long-only account a single ordinary position is capped at a total loss of what was risked in it, so ruin is normally reached by a sequence of large fractional bets compounding down, or by leverage and overnight gaps, rather than by one bad trade.
At a glance: how ruin probability climbs with bet size
The relationship is easiest to feel in a table. The figures below are a textbook illustration of the direction of the effect, not a recommendation of any win rate or bet size and not a claim about any real strategy: they simply hold a modest positive edge fixed and vary only the fraction of the account put at risk on each trade, to show how the chance of eventually hitting the floor responds.
| Fraction of account risked per trade | Effect on the account’s path | Direction of the risk of ruin |
|---|---|---|
| Very small | Drawdowns stay shallow; the floor is far away | Low, the edge has room to compound |
| Moderate | Normal losing streaks produce meaningful drawdowns | Higher, but survivable for a real edge |
| Large | Ordinary streaks now cut deep into the base | High, even with positive expectancy |
| Too large | A routine run of losses reaches the floor | Approaching certainty, the edge never matters |
The shape is the point. Hold the edge constant and the risk of ruin is overwhelmingly a function of the row you are in, not of the win rate. The same strategy is safe in the top row and doomed in the bottom one, having changed nothing but its bet size. This is why practitioners treat position sizing, rather than signal selection, as the variable that most directly governs survival, an argument developed in risk per trade.
How it is measured
For the simplest case, a fixed bet on each trade with a known win rate and a fixed win/loss size, there is a closed-form risk-of-ruin formula descended directly from the gambler’s-ruin solution, expressed in terms of the edge and the number of betting units the bankroll is divided into. Its two headline implications are the ones already described: ruin falls as the edge improves, and it falls sharply as the account is divided into more (therefore smaller) units per bet. The formula is a useful teaching tool precisely because it makes the bankroll-in-units term so visible.
Real strategies rarely satisfy the formula’s clean assumptions, wins and losses vary in size, outcomes are not independent, volatility shifts regime, so in practice risk of ruin is more often estimated by Monte Carlo simulation: run the strategy’s trade distribution forward thousands of times in randomised order and count how often a path reaches the floor. Both approaches deliver the same qualitative verdict. The exact number depends on assumptions that are hard to pin down; the ranking of what matters, size first, does not.
Survival is the precondition for compounding
The reason risk of ruin deserves top billing is that compounding is multiplicative and multiplication has an absorbing point. A single return of negative one hundred percent ends the sequence permanently: no subsequent gain, however large, recovers from zero, because everything after is multiplied by nothing. An unleveraged long position cannot on its own deliver that negative one hundred percent on the whole account (only on the slice committed to it), so in a long-only book the absorbing point is normally approached by a run of large fractional losses stacking up, not by a single trade. Even short of literal zero, deep drawdowns are punishing in a way losses do not feel until you do the arithmetic: recovering a large percentage loss requires a far larger percentage gain, and the deeper the hole, the more brutally asymmetric the climb out becomes. This is the same volatility-drag logic that makes holding several uncorrelated positions and controlling per-trade risk matter so much.
So the order of operations is not negotiable: survive first, compound second. A strategy that never risks ruin can afford to wait out any losing streak and let its edge express itself over a long horizon. A strategy that courts ruin is betting that the streak that ends it never arrives, and over enough trades, unlikely streaks stop being unlikely. Keeping the risk of ruin low is not caution for its own sake; it is the thing that keeps the account in the game long enough for a real edge to become real money. The growth-optimal way to think about how large a fraction to bet, and why betting more than that reduces long-run growth rather than increasing it, is the subject of the Kelly criterion.
The limits and the honest caveats
Risk of ruin is a powerful frame, but it is only as good as its assumptions, and a few of them are routinely violated.
- The inputs are estimates, not constants. Win rate and payoff ratio are measured from history, and history is a small, possibly unrepresentative sample. If the future win rate is worse than the backtested one, the true risk of ruin is higher than the calculation says. Numbers derived from an over-fit or survivorship- biased record understate the danger, as why backtests lie and survivorship bias both spell out.
- Independence is assumed but rarely holds. The tidy formula treats each trade as independent. Real markets cluster: losses arrive together in bad regimes, and a book of correlated positions can move as one. Correlated losses lengthen the effective losing streak and push the real risk of ruin above the independent- trade estimate.
- Gaps and tail events break the loss cap. Risk of ruin usually assumes the loss on a trade is bounded by a planned exit. Overnight gaps, halts, and disorderly liquidations can carry a single position through its intended exit, so a “fixed” per-trade loss becomes larger than modelled, and a rare tail event can inflict a drawdown the simulation never sampled.
- The floor is partly psychological. The mathematical floor may be zero, but the practical one is wherever the operator abandons the strategy. A drawdown that is survivable on paper can be unsurvivable in practice if it is deep enough that discipline breaks and the plan is abandoned at the worst moment. Behavioural ruin arrives before mathematical ruin.
How a systematic process treats it
A rules-based, systematic approach treats keeping the risk of ruin low as a design constraint rather than an afterthought, and the single most effective way to do that is fractional, risk-first sizing. Shishin is non-advisory and never tells anyone how much to risk or size, but the sizing logic inside the system that produces its publicly paper-traded record is risk-first: the loss budgeted to any one position is a small fraction of capital rather than a bet chosen by conviction, and because that fraction is expressed as a share of the current account, the bet automatically shrinks as the account draws down. That is precisely the property the gambler’s-ruin mathematics rewards, smaller units mean more of them, which means a lower chance of the streak that reaches the floor. The exact fraction is a production parameter and is not published; the principle is the ordinary one, that survival is engineered in through size, not hoped for. The realised, independently attested record that this sizing produces is visible at the live attestation page.
So: what is risk of ruin?
Risk of ruin is the probability that variance, in the form of an ordinary run of losses, drains an account to a floor it cannot recover from before its edge can compound. It rises as the win rate and payoff ratio worsen, but it is dominated by position size, to the point that a genuinely positive-expectancy strategy can be nearly certain to blow up simply by betting too large. The gambler’s-ruin mathematics explains why: a finite bankroll facing repeated bets is undone by the size of each bet relative to the purse, not by the edge alone. The practical upshot is that survival is the precondition for compounding, and the most reliable way to keep the risk of ruin low is to keep each bet a small fraction of the account, so that no ordinary losing streak can ever reach the floor. The edge earns the returns; the sizing decides whether the account is still there to collect them.
Sources & further reading
The probability and money-management tradition this concept draws on:
- Feller, W. An Introduction to Probability Theory and Its Applications, Vol. 1: the classical gambler’s-ruin problem and the mathematics of an absorbing barrier.
- Vince, R. The Mathematics of Money Management: fractional position sizing, drawdown, and the behaviour of risking a constant fraction of capital per trade.
- Vince, R. The Leverage Space Trading Model: extends the fractional-sizing tradition to multi-position portfolios and the Monte Carlo estimation of drawdown and ruin probability when trades are not independent or identically distributed.
- Kelly, J. L. (1956). “A New Interpretation of Information Rate.” Bell System Technical Journal, 35(4), 917 to 926: the growth-optimal bet fraction that risk-of-ruin thinking sits beside.
- Related reading on this site: risk per trade (the sizing framework that keeps ruin low) and the Kelly criterion (how large a fraction growth theory says to bet, and why more is worse).