Skip to main content

Risk of Ruin

Updated Jun 23, 2026 at 8:47pm

Research Draft Medium 1,070 words

Risk of Ruin (RoR) is the probability that a string of losses depletes your trading capital down to a terminal threshold — total bankruptcy, or more usefully a pre-set max-drawdown stop beyond which you stop trading — given your edge (win rate and payoff) and, above all, your bet size per trade. It descends directly from the classical gambler's ruin problem (Pascal, Huygens, Bernoulli, 17th century): even a player with a positive expected value can be wiped out by an unlucky sequence if each wager is too large a fraction of the bankroll. The central, counterintuitive lesson is that a positive-edge system can still ruin you — survival is governed less by how often you win than by how much you risk per bet. Position sizing, not win rate, is the primary lever over RoR.

How it's calculated / formed

The simplest closed form is the even-money version (win and loss are equal size), where the only inputs are your statistical edge and the number of "units" your bankroll is divided into:

RoR = ((1 − Edge) / (1 + Edge)) ^ (Bankroll / Bet Size)

Here Edge for an even-money game is win probability − loss probability (a 55%/45% game has Edge = 0.10), and Bankroll / Bet Size is the number of betting units — i.e. how many full bets you can absorb before broke (betting-data-lab; The Arca Labs).

For trading, where wins and losses differ in size, the same structure is used with a payoff-aware edge expressed in R-multiples and the exponent set to the number of risk units:

RoR = ((1 − Edge) / (1 + Edge)) ^ N
Edge = (WinRate × AvgWin) − (LossRate × AvgLoss)      (per-trade expectancy, in R)
N    = Account Size / Dollar Risk Per Trade            (number of risk units)

A 55% win rate at a 1.5:1 reward-to-risk ratio gives Edge = (0.55 × 1.5) − (0.45 × 1.0) = +0.375R of expectancy per trade; a \$50k account risking 1% (\$500) per trade has N = 100 risk units, driving RoR to effectively zero (BacktestBase; QuantifiedStrategies). The takeaway from the exponent is the whole point: RoR falls exponentially as N rises, so cutting bet size is the most powerful control you have.

More elaborate continuous forms exist. Ralph Vince (Portfolio Management Formulas, 1990) and others give a normal-approximation version such as RoR ≈ e^(−2·A·Z / D²) for fixed sizing — where A is mean return per trade, Z the fraction of capital risked, and D the standard deviation of trade returns (earnForex). These are refinements of the same idea; the even-money form above is the one worth memorizing.

How it's used in practice

RoR is mainly used backwards — not to compute a precise probability, but to justify a safe per-trade risk cap. Because the exponent N scales with 1 / (risk per trade), dropping per-trade risk from 5% to 1% multiplies the number of survivable losses and collapses RoR toward zero. A widely cited worked illustration: a \$10,000 bankroll with a ~2.3% edge has roughly ~1% RoR at 1% bet size, ~25% at 5%, and ~50% at 10% — same edge, the only variable changed is bet size (betting-data-lab). This is the quantitative backbone of the standard "risk only 0.5–2% of capital per trade" rule that pervades professional trading: at those sizes a positive-edge system's RoR is negligible, so the dominant survival risk becomes the things the formula ignores (below), not bad luck.

It also connects sizing decisions to losing streaks and drawdown depth: at 1% risk, a brutal 20-trade losing run costs ~20% of capital; at 5%, the same run is account-threatening. RoR is thus best read as a streak-survival budget.

Relationship to Kelly and over-betting

Risk of Ruin and the Kelly criterion are two sides of one coin. Kelly answers "what fraction maximizes long-run growth?"; RoR answers "what fraction keeps me alive?". The two are linked by a hard, well-documented fact: betting more than the Kelly fraction increases risk of ruin — betting exactly 2× Kelly produces zero expected growth, and anything beyond that drives the bankroll toward zero regardless of how large the edge is (Wikipedia, Kelly criterion). Even at full Kelly, drawdowns are violent — full Kelly carries roughly a 50% chance of a 50% drawdown — which is why practitioners use fractional Kelly (half or quarter), cutting both volatility and RoR at the cost of some growth (Wikipedia; multiple Kelly references). The practical synthesis: Kelly sets the ceiling, RoR (and the 0.5–2% rule) sets a far more conservative floor most traders should actually operate at. See the sibling nodes Full vs. Fractional Kelly and Estimating Edge & Odds.

Strengths & limitations

The strength of RoR is conceptual clarity: it proves rigorously that over-sizing, not low win rate, is what blows up otherwise-profitable accounts, and it gives a clean rationale for small fixed-fractional risk.

The limitations are severe and must travel with any computed number. The closed-form RoR assumes (1) a fixed bet size, (2) independent, identically-distributed trades, and (3) a constant, known edge. Real trading violates all three: correlated positions (several open swing trades all long the same regime go down together, acting like one oversized bet), edge that decays or disappears when a setup stops working, fat-tailed and gapping returns that the normal-approximation forms badly underestimate, and edge inputs that are themselves estimated with error — and over-estimating edge causes over-betting, the exact path to ruin. The single most common misuse is therefore trusting a tiny computed RoR as a guarantee of safety. A calculated RoR ≈ 0 is a lower bound and an illustration of how sizing dominates survival — not a promise. Many serious treatments abandon the closed form entirely and use Monte Carlo simulation (e.g. 10,000 sampled equity paths) to capture streaks and non-normal outcomes (BacktestBase).

System relevance

This node is a quantitative cross-link, not a standalone signal. It supplies the why behind the corpus's sizing nodes: Fixed-Fractional Sizing (the 0.5–2% rule this formula justifies) and the Kelly Criterion parent (over-betting Kelly → rising RoR). For Delvantic's Augustus trade-setup agent, RoR is a guardrail input, not a green light: it argues for conservative per-trade risk and for treating correlated open swing positions as a combined bet when sizing. Any RoR figure surfaced to a decision should be tagged as an idealized lower bound, with the iid/fixed-edge caveats attached — never as a survival guarantee.

Sources