ATI Learning / Performance Metrics
Maximum Drawdown in Trading: How to Read the Real Risk
Maximum drawdown measures the deepest observed fall from a previous peak. To understand the risk, you also need the curve, observation frequency, duration, cause and method of recovery.
The direct answer
What is maximum drawdown in trading?
Maximum drawdown is the largest observed decline in an account, portfolio or strategy from a previous high-water mark to a later low point during a stated period. If equity peaks at $12,500 and then falls to $9,000, the drawdown is 28%. It remains part of the record even if the account later recovers.
The percentage does not tell you the worst loss that could happen next. It describes the deepest decline found in the available sample, using a particular curve and observation method. A longer record, more frequent sampling or a future market shock can produce a deeper result.
Maximum drawdown is a historical path statistic—not a loss guarantee, safety rating or future ceiling.
For a complete system review, use ATI’s framework to evaluate an automated trading system. Our guides to algorithmic trading risks, trading bot warning signs and backtest versus live results provide the surrounding evidence checks.
Calculation
Measure from a running peak to a later trough
The order matters. A low that occurs before the peak cannot form a drawdown from that peak.
(Running Peak − Current Value) ÷ Running Peak
The largest observed drawdown in the period
No drawdown yet
New high-water mark
$3,500 below peak
Back to prior peak
$12,500 − $9,000 = $3,500 decline
$3,500 ÷ $12,500 = 0.28
0.28 × 100 = 28% maximum drawdown
The ATI Drawdown Profile
Five dimensions turn one percentage into usable evidence
No dimension replaces the others. A shallow balance decline can coexist with severe open-equity risk; a recovered drawdown can still reveal a fragile mechanism.
Basis
Which curve and sampling frequency created the number?
Balance · Equity · Intraday · End of dayDepth
How far did value fall in currency and percentage terms?
Peak · Trough · Capital remainingDuration
How long did the fall and total underwater period last?
Peak to trough · Trough to recoveryMechanism
Which trades, exposure and conditions produced the decline?
Losses · Leverage · Concentration · GapsRecovery
Did the same system recover without new capital or greater risk?
Version · Deposits · Position sizing · TimeDo not compare “maximum drawdown” figures until basis, period and observation frequency match. Then compare depth, duration, mechanism and recovery.
The most important distinction
Balance drawdown can miss open losses
Balance and equity answer different questions. Neither label should be omitted from a performance claim.
$10,000 peak → $9,200 low
The account records only completed trades and transactions. Floating profit or loss remains outside the balance until positions close.
$10,000 peak → $6,100 low
Equity includes the effect of open positions. It shows how much account value remained at the worst observed point.
Suppose losing positions remain open while profitable positions close. The balance curve can rise or decline modestly even while floating losses push equity much lower. If those positions later recover, the balance report may never record the depth of capital risk carried during the episode.
The 8% balance calculation is not necessarily false. It is incomplete if presented as the account’s full risk without the 39% equity decline. Myfxbook’s educational material distinguishes closed balance from equity that incorporates open profit and loss.3 MetaTrader likewise reports separate balance and equity drawdown measures.1
Observation frequency also matters. A daily equity snapshot may miss an intraday trough. TradingView’s intrabar calculation illustrates how bar highs and lows can be incorporated while a position is open.2
Calculation conventions
The same label can describe different measurements
Platform terminology is not universal. Define the series and formula rather than trusting a label such as “absolute” or “relative.”
| Convention | What it measures | What can be missed | Question to ask |
|---|---|---|---|
| Currency drawdown | Peak minus trough in account currency | Comparability across account sizes | What percentage of the peak was lost? |
| Percentage drawdown | Peak-to-trough decline divided by the peak | The dollar loss and account capacity | Which peak and curve were used? |
| Absolute drawdown | Often the fall below initial deposit on MetaTrader | Large losses from later, higher peaks | Does “absolute” mean below starting capital here? |
| Balance drawdown | Decline in closed account balance | Floating losses in open trades | What was worst equity at the same time? |
| Equity drawdown | Decline after open P&L is included | Brief troughs if sampled infrequently | Was equity sampled tick, bar, daily or on demand? |
| End-of-day drawdown | Decline using daily closing snapshots | Intraday stress and margin proximity | How deep was the worst intraday value? |
| Strategy drawdown | P&L path attributed to one strategy | Other bots and account-wide exposure | What was total account equity drawdown? |
| Prop loss limit | A contractual static or trailing breach threshold | Historical performance path | Is this a rule limit or an observed drawdown? |
Deposits can make account value jump and withdrawals can make it fall without representing trading performance. A credible analysis identifies cash flows and uses an appropriate adjusted return or growth series when measuring the strategy.
Recovery mathematics
Recovery becomes harder faster than the loss grows
The percentage gain is earned on the smaller amount of capital remaining after the decline.
Use the drawdown as a decimal: 0.30 ÷ 0.70 = 0.4286, or 42.86%.
| Drawdown | Capital remaining | Gain needed to recover |
|---|---|---|
| 10% | 90% | 11.11% |
| 20% | 80% | 25% |
| 30% | 70% | 42.86% |
| 40% | 60% | 66.67% |
| 50% | 50% | 100% |
| 60% | 40% | 150% |
| 75% | 25% | 300% |
| 90% | 10% | 900% |
The missing time dimension
Depth, decline time and time under water are different risks
Two 20% drawdowns can create very different capital, operational and behavioral demands.
Depth
The distance from peak to trough.
How much capital was lost?Peak-to-trough time
The time taken to reach the low.
How quickly did the failure develop?Recovery time
The time from trough back to the old peak.
How long did rebuilding take?Time under water
The full period from peak until a new high.
How long was capital below its high?Recovers in 5 weeks. Sharp gap or execution risk may dominate.
Recovers after another 18 months. Model decay and prolonged capital lock-up may dominate.
Maximum drawdown reports only the deepest observed decline. Average drawdown, drawdown frequency, duration distribution and longest time under water show whether the maximum was an isolated shock or one part of a repeated pattern.
Historical evidence has boundaries
Why the backtested maximum is not a future limit
A maximum grows as more observations and different conditions enter the sample. The next path can exceed the deepest path already seen.
Sample length
A short record has fewer opportunities to encounter an extreme sequence.
Selection
The published strategy may be the best survivor among many tested alternatives.
Market change
Future volatility, correlation and liquidity can move outside the historical range.
Execution
Slippage, gaps, rejected orders and market impact can deepen live losses.
Model assumptions
Daily or bar-level data may miss intraperiod equity extremes.
Scaling
Larger positions, leverage and portfolio correlation can change the path.
NFA notes that hypothetical trading cannot fully represent the effects of liquidity, slippage and financial risk.5 FCA guidance emphasizes testing across stressed scenarios rather than relying on one historical sample.6
Use a historical maximum as a reference for stress design, not as the stop point at which risk suddenly begins.
Look beneath the percentage
What caused the drawdown?
The same depth can come from an ordinary losing sequence or a mechanism capable of accelerating toward ruin.
Edge variance
Losses occur within stated sizing and stop rules. Exposure remains bounded.
Inspect trade distribution and recovery assumptions.Correlated exposure
Several positions behave like one large bet during stress.
Inspect account-wide exposure, not each trade alone.Loss amplification
Borrowing or derivatives magnify moves and can trigger forced liquidation.
Inspect margin use and liquidation distance.Averaging or martingale
Exposure increases while the market moves against the strategy.
Inspect the full position sequence and worst floating loss.Gaps and failed exits
Stops fill worse than expected or orders do not execute.
Inspect order logs, slippage and incident reports.Regime or edge decay
The relationship supporting the strategy weakens or disappears.
Inspect same-period backtests and retirement rules.The question without a universal answer
What is a good maximum drawdown?
No percentage is automatically good. A credible answer depends on the return objective, leverage, strategy, liquidity, evidence quality, capital constraints and the loss the user can financially tolerate.
Is it measured completely?
Prefer observed equity with clear frequency, not an unlabeled balance number.
Is return comparable?
Use the same period, capital convention and net-of-cost basis.
Is the mechanism bounded?
Identify leverage, concentration and whether exposure grows during losses.
Is the record long enough?
More trades and regimes strengthen context but never create certainty.
Can the account survive worse?
Risk limits need room for outcomes beyond the historical maximum.
Can the user tolerate the duration?
A sustainable system must fit capital needs and ability to remain invested.
Ratios such as annualized return divided by maximum drawdown can aid comparison when periods and formulas match. They remain sensitive to one historical trough, short records and unusually favorable years. Use them as supporting measures—not grades.
Worked comparison
Which system has lower drawdown risk?
The headline figures point one way. The complete evidence points another.
| Field | System North | System South |
|---|---|---|
| Stated annualized return | 24% | 38% |
| Reported maximum drawdown | 13% | 9% |
| Measurement basis | Intraday equity, net costs | Closed balance only |
| Worst observed equity drawdown | 13% | 34% |
| Track record | 36 months / 640 trades | 9 months / 115 trades |
| Longest time under water | 8 months | Not disclosed |
| Recovery evidence | Same version, no deposits | Open losses recovered; version then changed |
System North has the more complete, comparable risk record. System South’s advertised 9% balance drawdown does not support the implication that it experienced less capital risk. That does not prove fraud or make North suitable; it means the first comparison was not like for like.
Before trusting the number
Eight drawdown questions for a trading-system provider
- 01
Is the figure calculated from balance, equity or another series?
- 02
How frequently was equity observed, and were intraday open losses captured?
- 03
What were the exact peak, trough, dates, percentage and currency loss?
- 04
How long did peak-to-trough and full recovery take?
- 05
Which positions, leverage and sizing changes caused the maximum?
- 06
Did deposits, withdrawals, credits or fees affect the curve?
- 07
Did the same system version and risk settings produce the recovery?
- 08
Which controls activate before live drawdown exceeds the historical result?
Questions and answers
Maximum drawdown FAQs
How is maximum drawdown calculated?
Track the running peak of the chosen value series. At every later observation, calculate (peak minus current value) divided by the peak. The largest observed percentage decline in the stated period is maximum drawdown.
Should maximum drawdown use balance or equity?
For capital-risk evaluation, equity is generally more complete because it incorporates open profit and loss. Balance remains useful for closed-trade analysis. A credible report labels both and states the equity observation frequency.
What is a good maximum drawdown in trading?
There is no universal threshold. Judge the figure against its measurement method, return, leverage, duration, strategy mechanism, evidence length, capital limits and capacity to survive a worse future path.
Is maximum drawdown the same as a losing streak?
No. A losing streak counts consecutive losing trades. Drawdown measures the cumulative fall from a running account peak, which can include several loss-and-win sequences, open positions, costs and overlapping exposure.
Why does a 50% drawdown require a 100% gain?
A 50% decline leaves half the original peak. Returning from 50 to 100 requires doubling the remaining amount, which is a 100% gain. The required recovery formula is drawdown divided by one minus drawdown.
Can maximum drawdown exceed 100%?
For an unleveraged account limited to losing its capital, account equity ordinarily bottoms at zero, a 100% loss from the peak. With borrowing, derivatives or liabilities, losses can exceed deposited capital, so the economic loss can go beyond the original investment.
Does historical maximum drawdown predict the next drawdown?
No. It provides a historical reference. Future depth depends on the strategy, sample, market conditions, costs, leverage and controls. Stress tests and live monitoring should consider outcomes beyond the observed maximum.
What is the difference between maximum drawdown and a prop-firm drawdown limit?
Maximum drawdown is an observed performance statistic. A prop-firm limit is a contractual breach threshold, often calculated from starting balance, daily equity or a trailing high. Its formula must be read from the specific rulebook.
The ATI principle
A drawdown number is only as honest as the curve behind it
Begin with equity, but do not stop at depth. Record when the decline occurred, how long capital remained below its peak, which exposure created the loss and how recovery was achieved.
The goal is not to find a system that never draws down. It is to determine whether the observed risk is measured completely, structurally explainable and survivable under conditions worse than the historical sample.