What Risk of Ruin Is

You’ve learned to find an edge (Kelly) and to stress-test it across thousands of simulated futures (Monte Carlo). Now meet the constraint that quietly governs both: survival. Compounding is a privilege reserved for accounts that are still open. A strategy can have a real edge, a gorgeous long-run growth rate, and a beautiful backtest — and still hand its owner a coin-flip chance of going broke before any of that matters. This lesson defines ruin precisely, recaps the gambler’s ruin result that even a perfectly fair game bankrupts a finite bankroll, and confronts the paradox that anchors the entire topic: a winning system can still ruin you.

Ruin is an absorbing barrier

Analogy. Think of your account as a player walking along a cliff edge. Each trade nudges them left or right. Most days they wander harmlessly. But there is a drop behind them — and the cliff has a special, cruel property: once you fall off, you do not climb back. The market keeps moving, the edge is still there, but you are no longer in the game. Ruin is that cliff: a level your equity can reach but never leave.

Definition. Ruin is hitting an absorbing barrier — a state the process can enter but never exit. In trading it means your capital falls to a point where you can no longer continue: a margin call that closes you out, a fund whose investors all redeem after a 50% loss, a prop account that breaches its drawdown limit, or literally zero. The key word is absorbing: at the ruin level, the game is over for you regardless of what happens next.

Risk of ruin is then simply the probability of ever touching that barrier — over a horizon, or over an infinite future. It is a number between 0 and 1, and the whole topic is about computing it, lowering it, and not fooling yourself about it.

Two features make ruin different from an ordinary loss:

• It is path-dependent, not endpoint-dependent. A drawdown that you recover from is painful but survivable. Ruin cares only about whether you touch the barrier along the way — the order and timing of losses, not just where you end up.
• It is irreversible. A 50% loss needs a 100% gain to undo. A 90% loss needs a 900% gain. Below some point, the arithmetic of recovery becomes so brutal that the barrier is effectively absorbing even if the literal balance isn’t zero.

Gambler’s ruin: even a fair game bankrupts a finite purse

Before we get to winning systems, absorb the result for a perfectly fair one — because it’s more unforgiving than intuition suggests.

Setup. You start with a stake of $a$ dollars and play a fair game: each round you win or lose $1 with equal probability (a driftless ±1 random walk). You stop only when you either reach a target$b > a$or hit$0$ (ruin). What’s the chance you’re ruined first?

The classic result. For a fair game, the probability of ruin before reaching the target is $P(\text{ruin}) = \frac{b - a}{b} = 1 - \frac{a}{b}.$ Your survival probability is just $a/b$ — the fraction of the way to the target your starting stake represents.

Worked example. Start with $a = 20$ aiming for $b = 100$: $P(\text{ruin}) = \frac{100 - 20}{100} = \frac{80}{100} = 0.80.$ An 80% chance of going broke before you ever see $100 — in a game with zero house edge. Fairness protects your expected bankroll (it never trends down), but it does nothing for your survival, because the √t wandering of a random walk will touch the zero wall four times out of five on the way to a distant target.

The infinite-house limit. Now let the target recede to infinity ($b \to \infty$), which models playing forever against a house with effectively unlimited money: $P(\text{ruin}) = 1 - \frac{a}{b} \;\longrightarrow\; 1.$ Against an infinite bankroll, ruin in a fair game is certain. You cannot out-wait variance. This is the hard core of gambler’s ruin: a finite purse versus an unbounded opponent loses with probability 1, even when every single bet is fair.

The central paradox: a winning system can still ruin you

If a fair game can bankrupt you, what about a winning one? Surely a positive edge saves you? Not by itself — and this is the single most important idea in the topic.

The claim. A system with positive expectancy (it makes money on average per trade) can still have a substantial risk of ruin. Edge lowers ruin; it does not eliminate it. Whether you survive depends on three things working together — your edge, your bet size, and the path of wins and losses — not on edge alone.

Why. Expectancy is an average over many trades. Any particular stretch can deviate wildly from that average: a 55%-win system will still throw the occasional run of eight or ten straight losses (it’s not even rare — over hundreds of trades it’s nearly guaranteed). If each loss takes a big enough bite, that ordinary streak drives equity into the absorbing barrier before the edge has a chance to reassert itself. The edge is real; it’s just slower than variance in the short run.

Worked example — the over-bettor. Take a genuinely favorable even-money bet: win probability $p = 0.55$, so the per-bet expectancy is $E = (0.55)(+1) + (0.45)(-1) = +0.10 \text{ per unit staked} > 0.$ A clear edge. Now bet a large fraction — say 30% of your bankroll each time. A run of just a few losses compounds savagely: after four straight losses you hold $0.70^4 \approx 0.24$ of your bankroll — a 76% drawdown — and a system that needs to win 55% of the time will absolutely produce four straight losses. The edge says “play this game”; the bet size says “but not like that.” Same edge, wildly different survival.

Let’s watch it in a simulator. Below, every path is the same winning system; only the bet size changes. Crank the risk slider up and watch a profitable edge get buried under a pile of paths smashing into the ruin floor.

When does this matter?

When to think hard about ruin

• Whenever you compound — i.e. bet a fraction of a moving bankroll. Ruin risk is baked into the multiplication.
• Whenever there’s leverage or a hard drawdown limit — prop firms, margin accounts, funds with redemption triggers. The barrier sits high and is easy to hit.
• Whenever the horizon is long. Ruin probability rises with the number of bets; a tiny per-trade ruin risk compounds over thousands of trades.

When it matters less

• A single, un-leveraged, buy-and-hold position in a diversified index has essentially no absorbing barrier short of the asset going to zero — there’s nothing forcing you to sell at the bottom, so a drawdown is not ruin unless you make it one.
• Tiny bet fractions push ruin probability so close to zero it’s negligible — which is exactly the lever the rest of this topic teaches you to pull.

Putting it together

Ruin is hitting an absorbing barrier — a capital level you cannot trade back from — and risk of ruin is the probability of ever touching it. It’s path-dependent and irreversible, and the barrier often sits well above zero (a redemption-triggering drawdown, a prop firm’s limit). Gambler’s ruin shows the math is harsher than intuition: a fair game still bankrupts a finite purse aiming at a distant target — ruin probability $(b-a)/b$ — and against an infinite house, ruin is certain. The central paradox is the load-bearing idea: a positive-expectancy system can still ruin you, because expectancy is a long-run average while ruin is decided by bet size and the path of wins and losses in the short run. Edge says the game is worth playing; risk of ruin says whether you’ll survive to collect. The rest of this topic makes both precise — and teaches you to keep the second one small.

Next up — expectancy and the edge — we open the per-trade engine: how to compute the average profit a system makes per trade, the elegant R-multiple bookkeeping that puts every trade on a single ruler of risk, and the surprising trade-off between how often you win and how much you win when you do.