A drawdown measures how far a trading account has fallen from its previous peak. A losing streak counts consecutive losing trades. Recovery mathematics shows the return needed to get back to that peak. These measures describe different parts of the same problem: losses reduce both your capital and the base from which future returns compound.
A 20% loss needs a 25% gain to recover. A 50% loss needs 100%. That asymmetry makes controlling losses central to speculative trading. The objective is not to avoid every losing trade, but to prevent a difficult period from creating a recovery requirement the account cannot realistically support.
How to Calculate Trading Drawdown
For an account with no deposits or withdrawals during the measurement period, calculate percentage drawdown as:
Drawdown (%) = (Previous equity peak − Current equity) ÷ Previous equity peak × 100
Suppose an account starts at £10,000, rises to £12,000 and then falls to £9,600. The drawdown is £2,400, or 20% of the £12,000 peak. Measuring the fall against the original £10,000 deposit would answer a different question. The account has lost 4% of its starting capital, but it has fallen 20% from its best level.
The reference peak, sometimes called the high-water mark, changes only when equity reaches a new high. If the account recovers to £11,000, the reference remains £12,000 and the current drawdown is approximately 8.33%.
Current Drawdown Versus Maximum Drawdown
Current drawdown describes the account’s position now. Maximum drawdown is the largest percentage fall from a peak to a subsequent trough within the period examined. A later recovery reduces current drawdown but does not erase the historical maximum.
Measurement frequency matters. A record using only month-end values can miss a sharp fall and recovery within a month. CME Group’s discussion of drawdown depth and reporting frequency highlights why monthly performance figures may conceal larger losses between reporting dates.
For account monitoring, use equity that includes unrealised gains and losses, not just the balance after closed trades. An account with £10,000 in cash balance and £2,000 of open losses has £8,000 of equity before further charges. Ignoring those open losses does not make the drawdown disappear.
Keep deposits and withdrawals separate from trading performance. Adding £2,000 to a damaged account increases its equity, but it does not represent a trading recovery. Use a cash-flow-adjusted performance series, such as a unitised account value, when measuring drawdown across funding changes.
Why Recovering a Loss Requires a Larger Gain
The recovery percentage is larger because the gain is earned on a smaller amount of capital. After a £10,000 account loses 20%, it holds £8,000. Recovering the missing £2,000 requires a return of £2,000 ÷ £8,000, which is 25%.
If the drawdown is expressed as a decimal, the formula is:
Required recovery return = Drawdown ÷ (1 − Drawdown)
| Drawdown | Remaining equity | Gain needed to recover |
|---|---|---|
| 5% | £9,500 | 5.26% |
| 10% | £9,000 | 11.11% |
| 20% | £8,000 | 25.00% |
| 30% | £7,000 | 42.86% |
| 40% | £6,000 | 66.67% |
| 50% | £5,000 | 100.00% |
| 75% | £2,500 | 300.00% |
These calculations assume no further deposits or withdrawals. The recovery returns must be achieved after trading costs. Restoring the old nominal account value also does not compensate for inflation or the time spent recovering.
A gain and loss of the same percentage do not cancel out. Losing 20% and then gaining 20% turns £10,000 into £9,600. The order of those two returns does not change their compounded result.
Losing Streaks Do Not Tell You the Whole Story
A losing streak counts consecutive losses, not their size. Five losses of £20 cause less damage than two losses of £500. A drawdown can also continue through several winning trades if those gains fail to restore the previous equity peak.
Win rate alone therefore cannot describe drawdown risk. Average win size, average loss size, position sizing, costs and the order of results all matter. The relationship between profitability and survival is covered in trading expectancy and risk of ruin; here, the concern is how losses can accumulate along the way.
The Probability of Consecutive Losses
Under a simplified model where trades are independent and the probability of losing remains constant, the probability that the next n trades all lose is:
Probability of n consecutive losses = Loss probabilityn
With a 50% loss probability, the chance that the next five trades all lose is 0.55, or 3.125%. That is not the probability of encountering at least one five-loss streak somewhere within 100 trades. A longer sequence offers many more opportunities for a streak to occur.
To illustrate the difference, divide 100 independent trades into 20 separate blocks of five. The probability that at least one complete block contains five losses is:
1 − (1 − 0.55)20 ≈ 47.0%
This is a lower bound for the chance of a five-loss streak anywhere in the 100 trades. It excludes streaks crossing the block boundaries. Calculating the full probability requires accounting for overlapping runs; simply multiplying 3.125% by the number of possible starting points gives the wrong answer.
The independence assumption also needs scrutiny. If several positions express the same market view, treating their outcomes as independent may be inappropriate. Use the formula as a model with stated assumptions, not as a promise about the next trading session.
Under that independent model, five previous losses do not make the next trade more likely to win. The market does not owe the account a refund.
How Position Sizing Changes Drawdown Depth
Consider a simplified account that loses exactly the same fraction of its remaining equity on every losing trade. With no intervening wins or cash flows:
Remaining equity = Starting equity × (1 − Risk fraction)Number of losses
Ten consecutive losses reduce equity by approximately 9.56% at 1% risk per trade, 18.29% at 2%, and 40.13% at 5%. These are calculated scenarios, not estimates of how often such streaks occur.
With a fixed percentage approach, the cash amount at risk falls as equity falls. CME Group’s comparison of percentage risk rules demonstrates this shrinking cash exposure during a losing streak. It slows depletion, though it cannot turn an unprofitable strategy into a profitable one.
Keeping cash risk unchanged produces a different result. Risking £200 per trade from a £10,000 account initially means risking 2%. After ten losses of £200, equity is £8,000 and the same £200 represents 2.5%. The nominal stake has not increased, but its burden on the account has.
The compounding formula assumes each loss equals the stated fraction. Actual losses can exceed the amount planned. For stock stop orders, the trigger price is not a guaranteed execution price, and a stop-limit order may remain unfilled. The SEC investor bulletin on stop orders sets out that distinction.
Practical sizing also has constraints. Contract sizes may prevent an exact percentage allocation, while simultaneous positions can create more account exposure than a per-trade figure suggests. A rule that sounds conservative in isolation may be much less conservative when applied to several positions at once.
There is no universally safe percentage. Set exposure through a position sizing and trading risk budget that considers the account’s capacity for loss, combined positions and adverse execution. The percentages above illustrate the mathematics, not a recommendation.
Drawdown Duration and Time to Recovery
Depth measures how much the account has lost. Duration measures how long it remains below its previous peak. An account down 8% for eighteen months presents a different practical problem from one that falls 8% and recovers within several weeks.
Define the time measurements clearly. Peak-to-trough time runs from the previous high to the eventual low. Recovery time runs from that low back to the previous high. Total time underwater covers both. If recovery has not happened, the underwater period remains open; its final length is unknown.
It is tempting to estimate recovery time by dividing the required gain by an average return. Compounding makes that unreliable even before allowing for uneven results. After a 20% drawdown, a hypothetical uninterrupted return of 1% per month would require 23 whole months to exceed the old peak:
0.80 × 1.0123 ≈ 1.006
That is an arithmetic illustration, not a forecast. Realised returns need not arrive steadily, and further losses can delay or prevent recovery. Reducing exposure may slow a subsequent recovery if the strategy becomes profitable again, but protecting remaining capital can still justify that decision.
Why Historical Maximum Drawdown Is Not a Safety Limit
A backtest with a 12% maximum drawdown establishes what happened in that simulation. It does not establish that future losses cannot exceed 12%. The result belongs to a particular dataset, trading rule, cost assumption and sizing method.
Repeatedly adjusting a strategy until its historical results look attractive creates another problem: the selected version may fit past noise rather than a repeatable opportunity. The research paper The Probability of Backtest Overfitting examines how testing and selecting strategies on historical data can produce misleading performance results.
For drawdown planning, examine more than the single worst historical observation. Compare different periods, test less favourable costs and consider outcomes with smaller wins or larger losses. Those exercises ask whether the account could tolerate results worse than the original test.
Simulation can help frame those questions, provided its assumptions remain visible. Shuffling a fixed list of trade returns explores different orders of the same outcomes. It cannot create a loss larger than anything in that list. Sampling individual trades also removes their original sequence, which may discard loss clustering.
Do not read a simulated percentile as a guarantee. A modelled 95th-percentile drawdown means 95% of that model’s simulated outcomes fell at or below that level. It does not establish 95% real-world protection against exceeding it.
Build a Drawdown Response Before Losses Arrive
A drawdown plan should connect observable conditions to decisions. Avoid rules that depend on feeling calm after a bad session. Define what you will measure, when you will review it and what would justify reducing or stopping exposure.
- Measurement: Track current equity, the adjusted equity peak, drawdown percentage and time underwater.
- Review triggers: Specify when losses, execution problems or departures from the strategy require investigation.
- Exposure rules: Decide in advance whether a trigger means smaller positions, no new trades or a pause.
- Restart conditions: Require a documented review and evidence that the original problem has been addressed.
Keep account-level drawdown triggers separate from individual trade stops. One concerns the trading programme; the other concerns an open position. Neither should depend solely on whether the latest trade won or lost.
During a review, separate correctly executed losing trades from rule violations and execution failures. A loss within the tested strategy calls for a different investigation from an oversized position or an order placed by mistake. Avoid changing every rule after a short run, but do not dismiss every deterioration as bad luck either.
Increasing stakes simply to recover the old peak replaces a risk decision with an account-balance target. That belongs to the pattern of revenge trading and loss-chasing, not a recovery plan.
The old peak is useful for measurement. It is not a deadline, and it is not a reason to take a trade. Recovery remains uncertain; controlling the exposure committed to pursuing it is the decision still available.