WhitmanTrading

Risk of Ruin Calculator

Risk of ruin is the probability that a sequence of losses reduces an account to nothing before an edge has time to show. It falls exponentially as position size falls, so halving what you risk per trade cuts the probability by far more than half.

The chance the account does not survive

Defaults use a win rate of 52% with even-money outcomes, risking 2% of the account per trade.

Chance of ruin 1.83%
Losing trades that would empty it 50
Chance at half this risk 0.0334%
Chance at double this risk 13.52%

This is the classic gambler’s-ruin form and it assumes even-money outcomes, a fixed stake, and independent trades. Real trades are none of those things, so read the answer as a comparison between position sizes rather than as a forecast of your account.

Runs entirely in your browser. Nothing you type is sent anywhere or stored.

How the number is built

A candlestick chart with a series of losing trades marked.
The chance a losing run ends the account. Illustrative chart - not real market data.

Two inputs and one exponent. The exponent is where all the behaviour lives.

Ruin = ((1 − W) ÷ W) ^ (100 ÷ risk %)

where W is the win rate and the exponent is how many full-size losses the account can absorb.

The first half of a price series with wins and losses marked.
A win rate, a stake, and how many stakes fit in the account. Illustrative chart - not real market data.

The base of that power is below one whenever you have an edge, so raising it to a larger exponent drives it toward zero fast. Larger positions mean a smaller exponent, and a smaller exponent means the number barely falls at all.

A worked example

Take the defaults: a win rate of 52% with even-money outcomes, risking 2% per trade.

The account absorbs 100 ÷ 2 = 50 consecutive full-size losses.

The base is 0.48 ÷ 0.52 = 0.9231.

And 0.9231 raised to the 50th power is 1.83%.

The second half of a price series with positions scaled differently.
Position size moves the answer more than the win rate does. Illustrative chart - not real market data.

Now change only the risk. At 1% per trade the same edge gives 0.0334%. At 4% it gives 13.52%. Doubling the position size multiplied the danger by about seven, and halving it divided the danger by about fifty-five.

The win rate matters less than it feels

A window of price bars where a small advantage accumulates.
Without an edge, ruin is certain given enough attempts. Illustrative chart - not real market data.

Hold risk at 2% and vary the win rate instead. At 50% ruin is 100% — certain, eventually, at any position size. At 51% it is 13.53%. At 52% it is 1.83%. At 55% it is 0.0044%.

Both inputs matter enormously and only one of them is under your control this afternoon. You cannot decide to win more often; you can decide to risk less.

A long-horizon candlestick view of small consistent positions.
Small positions survive long enough for an edge to appear. Illustrative chart - not real market data.

Which is the entire practical message of this page. An edge only pays if the account is still there when the sample gets large enough for the edge to show, and position size decides that.

Losing runs are normal

A section of the price series with a run of same-direction bars.
Runs of the same outcome are the ordinary case. Illustrative chart - not real market data.

On this site’s shared series the average direction run is 2.01 bars, measured across 286 runs. A run of two is unremarkable and expected.

A candlestick series with an extended one-directional stretch.
And the longest run measured eleven. Illustrative chart - not real market data.

The longest run in that same series was 11. A strategy sized so that eleven losses would be serious is a strategy sized for a sequence that has already occurred once in 576 bars — and a real trading record is nothing like the neat sequence this formula assumes.

What the formula cannot see

A candlestick chart annotated with the round-trip cost of a switch.
Costs come out of the edge, which is the base of the exponent. Illustrative chart - not real market data.

Costs reduce the win rate before this calculation starts. On this site’s shared series a round trip measures about 2% of the median bar range of 0.493 — so a win rate of 52% measured before costs may be 50% after them, and 50% is where the answer becomes 100%. The figures are in research/series-measurements.json.

A candlestick chart with a volume histogram beneath it.
And a loss can exceed the stake you chose. Illustrative chart - not real market data.

The derivation assumes a loss is exactly the size you risked. Slippage in a thin market makes it larger, and the exponent — the number of losses the account can take — falls with it.

Turning the answer into a size

Run it backwards and it becomes a sizing rule rather than a diagnosis. Decide what probability of ruin you are willing to carry, then find the risk per trade that produces it.

On a win rate of 52% with even-money outcomes, accepting a one-in-a-thousand chance means risking about 1.16% per trade. Accepting one in a hundred means about 1.74%. Accepting one in ten means about 3.48%.

Those three tolerances span 2.3 percentage points of position size, which is the practical shape of an exponential relationship: the difference between a careful account and a fragile one is far smaller than it feels, and it lives entirely in the sizing.

The original data

Of the 24,971 unique videos in research/search-study-corpus.jsonl, 2 have an instruction-shaped title about risk of ruin, at a median of 23,415 views across 2 channels — and 0% are calculator-shaped. The Kelly criterion appears in 2 at 96,269, losing-streak odds in 1 at 24, and position sizing in 195 at 1,738. The counts come from site/rank_tools2.py.

A candlestick series with several gaps, the largest of them marked.
A gap can take more than the risk you set. Illustrative chart - not real market data.

Two videos on the probability of losing everything, against 541 on prop-firm payouts at a 12,950 median. The subject with the largest consequence has the least coverage in the corpus, which is a fair description of the whole risk-management category.

A stretch of price bars cut short at a decision point.
Five losses in a row. Increase size to recover? Illustrative chart - not real market data.

The answer to the question on that chart is that increasing size after losses raises the exponent’s denominator and therefore the probability of ruin. Going from 2% to 4% after a bad run takes this account from 1.83% to 13.52% — at the exact moment the equity is lowest and the sequence is already running against you. The arithmetic says the correct move after losses is to risk less, not more, because ruin is measured from where you are rather than from where you started.

When it fails

The failure that matters is a win rate that was true and stopped being true. The formula takes the edge as a permanent property, so it will keep returning a comforting 1.83% from a number measured last year while the strategy has quietly become a coin flip — and at a coin flip the honest answer is 100%. Nothing in the calculation can detect that, and recalculating from a rolling sample is the only defence.

The second failure is assuming trades are independent. Losses cluster because conditions cluster.

A third is using even-money assumptions on a strategy with uneven payoffs. This form does not model them.

A fourth is ignoring costs. They come out of the win rate, which is the base of the exponent.

A fifth is treating a small percentage as zero. 1.83% is roughly one account in fifty-five.

And a sixth is raising size after a losing run. That is the one action the arithmetic forbids.

Position sizing is where the risk-per-trade input is actually decided. Expectancy is the edge this formula assumes you have measured. And drawdown is what ruin looks like before it finishes.

What I actually do

The number that made me change how I size positions was not the ruin figure itself but the gap between the 2% row and the 4% row. I had assumed doubling the risk roughly doubled the danger. It multiplies it by about seven, on these inputs, and that is the whole argument for trading small enough to be bored.

— Michael Whitman

This page is educational, not financial advice. Test every idea on your own charts before risking money.