A run of losses can feel like evidence that a strategy has stopped working. It can also occur under a stable process by chance. The probability depends on what counts as a loss, how many trades you include, and whether outcomes are independent. It does not, by itself, tell you the probability of losing your account.
“Risk of ruin” needs a definition before it can be estimated. Ruin might mean a zero balance, a 20% drawdown, falling below a venue’s maintenance requirement, or being unable to follow a stated plan. Each is a different event. A losing-streak calculation answers a narrower question.
One five-loss block is not a whole trading horizon
Assume every trade independently loses with probability 40%, with the same probability on every trade. The chance that five losses fill one preselected five-trade block is 0.40⁵ = 1.024%. This is the chance for those five slots. It is not the chance of encountering five consecutive losses somewhere in a longer record.
For the longer-horizon calculation below, each trade is either a win or a loss, draws are excluded, and the 40% loss probability never changes. I count paths with no five-loss run, keeping track of the current run length from zero through four. A win resets the count; a fifth consecutive loss enters the event. This method accounts for overlapping possible starting points instead of treating each window as independent. A mathematical treatment of run probabilities in Bernoulli sequences is included in the sources.
| Trades in the horizon | Chance of at least one run of 5 losses |
|---|---|
| 5 | 1.02% |
| 10 | 4.10% |
| 20 | 10.01% |
| 50 | 25.65% |
| 100 | 45.91% |
| 250 | 79.17% |
These are exact results for the stated model, rounded to two decimals. The rise with horizon is why a streak that looks unusual in five trades is much less surprising when you inspect hundreds. The chart describes a losing-run event only; it is not a forecast, a strategy estimate, or an account-ruin probability.
Convert losses into an equity path
Probability and account impact are separate calculations. Consider a fictional $10,000 balance and a rule that risks 1% of current equity on each trade. If each losing trade costs exactly that planned amount, with no fees, slippage, gaps, or other changes, then five losses leave $10,000 × 0.99⁵ = $9,509.90, a 4.90% drawdown. Ten leave $10,000 × 0.99¹⁰ = $9,043.82, a 9.56% drawdown.
Those figures describe a sizing rule applied to an assumed sequence; the 40% loss probability did not determine the dollar loss. A fixed-dollar risk rule would produce a different path. Actual stops can slip, costs accumulate, and leveraged products can have margin rules that force a closeout before the balance reaches zero. The simple calculation models none of those mechanics.
Increasing size after each loss to “win it back” changes the exposure precisely when the account has less room for error. It does not improve the probability of the next trade. Set a maximum exposure and a pause or review condition in advance; then assess rule adherence, costs, market conditions and the quality of the evidence before changing a strategy.
What an actual ruin model must specify
To estimate a threshold-crossing probability, define the threshold and simulate full equity paths under an explicit outcome and sizing model. At minimum, record:
- Failure event: for example, equity below a specified amount or drawdown beyond a chosen limit. For leveraged futures, distinguish that from exchange margin closeout and read the contract rules.
- Trade outcomes: win/loss probability, payoff sizes, costs and how these inputs were estimated. A win rate alone is not enough.
- Dependence: whether outcomes are treated as independent or allowed to cluster. Adjacent trades can share the same market regime; shuffling them independently can erase that structure.
- Sizing and horizon: fixed dollars, a fraction of current equity, or another rule; the number of trades; and any deposits, withdrawals or stopping conditions.
- Model uncertainty: the sample length, parameter choices, alternative assumptions and sensitivity of the result.
Then simulate the sequence trade by trade, updating equity, size and costs, and record whether the defined threshold was reached. Report the assumptions next to the result. A precise percentage from a short, selected backtest can look more certain than its inputs justify. Resampling does not create market regimes that the source record never contained; the separate trade-resampling guide explains that limit.
The input called a “loss” needs care too. A trade with positive gross movement can be net negative after fees, spread, funding or borrow; a small positive trade can still fail to recover a prior drawdown. State whether the event is defined from gross outcomes or net account changes, and apply the same definition across the sample and the simulation. Otherwise the streak count and the equity path describe different things.
The 40% example is a chosen input, not something inferred from a few recent trades. If you estimate a win rate from historical results, the estimate itself is uncertain and can shift when the market regime changes. Compare a range of plausible probabilities and payoff distributions, preserve time order when dependence matters, and show how often each defined threshold is reached under each scenario. Such simulations describe the assumptions you supplied; they cannot establish that those assumptions will persist.
The CFTC’s virtual-currency advisory warns that leveraged futures can require additional margin or a position closeout when markets move against the account, and losses may exceed the initial investment. That warning is about futures and margin-account risk; specific contract and venue rules matter. It is not a probability estimate for an individual trader or a claim that every crypto product behaves the same way.
SuperQuant’s synthetic examples can demonstrate arithmetic, but they cannot estimate your chance of ruin or model a live venue’s liquidation path. For a bounded planning exercise, read the guides to position sizing, drawdown and recovery, and crypto liquidation. These calculations are educational, not a trading recommendation or personal financial advice.