Risk of ruin: the probability your account doesn't survive
A strategy that wins 55% of its trades still faces a 13.4% chance of ruin when it risks 10% per trade — and 0.004% at 2%. The gambler's ruin formula turns position size into a survival probability.

A probability, not a prediction
Risk of ruin is the probability that a run of losses takes an account below the level where trading can continue — zero, a broker's margin requirement, or a funded account's loss limit. It is not the chance of losing one trade. It is the chance that a losing run ends the game before an edge has time to show up.
The classic formula treats each trade as a bet that wins or loses one fixed unit. Risk of ruin equals (q / p) raised to the power N, when p is greater than q — and 1 when p is equal to or lower than q. Here p is the win probability, q is 1 minus p, and N is the number of capital units: the distance to the ruin line divided by the dollar risk per trade. A $10,000 account risking $500 per trade holds 20 units.
The formula is the gambler's ruin result from probability theory. Blaise Pascal posed the problem in a letter to Pierre de Fermat in 1656, Christiaan Huygens published a version in 1657, and Grinstead and Snell's Introduction to Probability works it out in section 12.2. The trading version only changes the names: the stake is the account, and the opponent is the market.
- Full Kelly10% of capital per trade
- Half Kelly5% of capital per trade
- Investopedia cap20% to 25% in one equity
One win probability, three position sizes
Take a $10,000 account where ruin means losing the whole $10,000, every trade wins or loses the same dollar amount, trades are independent, and the win probability is 55%, or 0.55, so q is 0.45 and q divided by p is 0.8182. The only thing that changes is the size.
Risk $1,000 per trade, or 10% of the account: N is 10, and 0.8182 to the power 10 is 0.134, a 13.4% chance of ruin. Risk $500, or 5%: N is 20, and the result is 1.8%. Risk $200, or 2%: N is 50, and the result is 0.004%. Cutting the risk per trade from $1,000 to $200 cuts the ruin probability from 13.4% to 0.004% — the same edge, the same account, a different size.
The result is just as sensitive to the edge itself. At a win probability of 52%, or 0.52, the same three sizes give 44.9%, 20.2% and 1.8%. A 3-point error in the estimated win probability multiplies the ruin figure more than tenfold at the $500 size. And at a win probability of 50%, or 0.50, the formula gives 1 for every size: with no edge and no end point, a long enough losing run eventually arrives, whatever the size.
Kelly says 10% — ruin math says 13.4%
The Kelly criterion answers a different question: what fraction of capital maximizes long-run growth. John L. Kelly Jr. published it in 1956 while working at Bell Labs, and Investopedia states the formula as K equals W minus (1 minus W) divided by R, where W is the win probability and R is the win-to-loss ratio.
Take W of 0.55 and R of 1, an even-money trade. K is 0.55 minus 0.45, or 0.10 — risk 10% of capital per trade. That is exactly the N equals 10 size from the section above, and the ruin formula prices that size at 13.4%. The size that maximizes growth still carries about a one-in-seven chance of ending the game. Half-Kelly, 5% per trade, is N equals 20, and the ruin figure falls to 1.8%.
Investopedia adds a rule of common sense regardless of what the Kelly percentage says: commit no more than 20% to 25% of capital to one equity. Scholars have noted that the Kelly criterion can be risky in the short term because it can indicate initial wagers that are significantly large, and it assumes the odds are known precisely — in real markets they never are. Kelly maximizes growth; ruin math measures survival. The two answers differ because they ask different questions.

What the model assumes — and what to do with it
The formula is exact only when its assumptions hold, and real trading breaks most of them: equal wins and losses, the same fixed dollar risk on every trade, independent trades, a known and constant win probability, and no end point. Read the result as the output of a model, not as a forecast. When wins and losses are unequal, the closed form no longer applies, and Investopedia notes that risk-of-ruin calculations can be done with a value-at-risk measure or through techniques like Monte Carlo simulation.
Three moves lower the number. First, risk less per trade: each step up in N cuts the result, and it falls fast as N rises. Second, check the win probability on enough trades: a figure measured over a few dozen trades is a weak input. Third, keep losses at the planned size — moving an exit or adding to a loser breaks the equal-loss assumption — and set a daily or weekly loss limit that ends a bad run before it reaches the account limit. CME Group's risk-management course makes the same point in plainer terms: know where you will exit before you enter, and let the distance to that exit set the size. Size is the one input fully under the trader's control. Educational content, not investment advice.



