Risk of ruin

2026-10-26

Risk of ruin is the probability that a sequence of trades takes your account below the level at which you can continue. It is the one risk number that combines edge, position size and endurance — and it is dominated by position size far more than by edge.

That last point is the useful one, and it is counterintuitive enough to be worth arriving at slowly.

The simplest honest version

Ruin formulas come in several shapes, most of which assume more than they can support. The version that is exactly true and requires nothing:

With a fixed fraction f risked per trade, after k consecutive losses the account is:

remaining = (1 − f)^k

And the probability of k consecutive losses at win rate p is (1 − p)^k.

That is enough to answer the practical question, and it does not pretend to know things it does not.

What position size does

Fix the win rate at 45% and vary only the risk per trade. Loss after a ten-trade losing streak — which, per losing streak probability, you should expect to see within a few hundred trades:

Risk per trade	After 10 losses	Recovery required
0.5%	−4.9%	+5.1%
1%	−9.6%	+10.6%
2%	−18.3%	+22.4%
5%	−40.1%	+67.0%
10%	−65.1%	+186.7%
20%	−89.3%	+831.4%

The edge is identical in every row. Only the sizing changed.

At 1% you have had a bad fortnight. At 10% you have an account that has to triple to recover. At 20% you are functionally finished. Same strategy, same win rate, same streak.

This is why risk of ruin is mostly a sizing question. A positive edge does not protect you from oversizing, because ruin arrives through variance and variance scales with f.

The formulas and what they assume

Published risk-of-ruin formulas typically assume:

A known, constant win rate. Yours is estimated from a limited sample drawn from market conditions that have since changed.
Independent trades. They are not — losses cluster, because the conditions producing one produce the next. This makes long streaks more likely than the arithmetic says.
A constant payoff. Realised R varies with slippage, partial fills and early exits.
Exact fractional sizing. Lot sizes, minimum notionals and margin tiers interfere.
No costs. Fees and funding come out of the same balance.

Every one of those assumptions fails in the direction that makes real risk higher than computed. So treat any ruin figure as a floor, and be suspicious of one quoted to two decimal places.

Ruin is not zero

A detail people miss: ruin does not mean a zero balance. It means falling below the level at which you can keep going, which is higher and more personal.

That might be the margin required for a normal position, the point at which the account is too small to be worth the attention, or the drawdown past which you stop trusting the approach and stop executing it properly. The last of those is the binding constraint for most people, and it arrives well before the arithmetic one.

Which means the number to protect is not the balance. It is the largest drawdown you would continue trading through — and that is worth deciding in advance, calmly, rather than discovering.

The more useful question

Instead of “what is my risk of ruin”, which requires inputs you do not reliably have, ask:

What drawdown would make me stop, and how many consecutive losses does it take to get there at my current size?

Both terms are knowable. Solve (1 − f)^k = 1 − D for k:

k = log(1 − D) ÷ log(1 − f)

At 2% risk with a 30% tolerance: log(0.70) ÷ log(0.98) ≈ 17.6 — about eighteen consecutive losses. Compare that to the expected longest streak for your win rate. If your expected streak is nine and your tolerance is eighteen, you have room. If they are the same number, you do not.

That comparison uses only things you can measure, and it produces an actionable answer: change f.

Ways ruin arrives faster than the arithmetic

Correlated positions. Three trades on correlated instruments is one trade at three times the size, and the sizing formula does not know that.

Leverage crossing liquidation. If liquidation sits inside your stop, a losing trade costs the margin rather than the amount you sized for — see the leverage and stop-loss table. One such trade can equal many normal losses.

Sizing up during a drawdown. The impulse to recover faster increases f exactly when the balance is smallest. This is the single most common path to ruin for accounts that had a real edge.

Costs. At high turnover, fees and funding erode the balance independently of whether trades win or lose.

What actually protects you

In descending order of how much they help:

A risk fraction small enough that your expected longest streak is survivable. This does most of the work.
A daily loss limit enforced in code, so one bad day cannot compound into the streak arithmetic above.
Position caps that do not move, so f cannot drift upward during a drawdown.
Awareness of correlation, so three positions are not secretly one.

Notice that none of these require improving your edge. Ruin is a variance problem, and variance is controlled by size.

FAQ

What is an acceptable risk of ruin?

Lower than the number you first think of, and the framing is more useful than the threshold. Because the assumptions behind any ruin calculation understate real risk — correlated trades, estimated win rates, costs — a figure that looks comfortably small on paper is optimistic in practice. The more robust approach is to check that your expected longest losing streak is comfortably survivable at your current sizing.

Does a higher win rate protect me from ruin?

Much less than reducing position size does. Compare the table above: the same win rate across every row, with outcomes ranging from a mild setback to an unrecoverable account, driven entirely by the risk fraction. Edge determines whether you make money over many trades; sizing determines whether you are still there to collect it.

Is risk of ruin the same as maximum drawdown?

No. Maximum drawdown is the largest peak-to-trough decline that happened; risk of ruin is the probability of a decline large enough to end the account. They are connected through the recovery arithmetic — a large maximum drawdown implies a recovery requirement that may not be realistic, which is the practical form ruin usually takes.

How do I calculate it for my own trading?

Rather than a formula requiring inputs you cannot verify, solve for how many consecutive losses reach your personal stopping point: k = log(1 − D) ÷ log(1 − f), where D is the drawdown you would quit at and f is your risk per trade. Compare that k against the expected longest streak for your win rate. If they are close, reduce f.