Max Drawdown

Max Drawdown captures the largest decline from an earlier peak to a subsequent lower value in the observed interval. It is not the difference between an arbitrary maximum and minimum regardless of their order. Each observation is compared with the highest value reached up to that point within the specified window. The deepest of these declines is the maximum drawdown. It may span several losing and winning steps; not every intermediate step needs to be negative. A positive start-to-end return therefore does not rule out a substantial decline within the period. [1]

This entry reports drawdown depth as a nonnegative percentage. If V denotes the current value and H the positive running peak, we use d = 1 - V / H and convert the result to a percentage. The maximum is taken over the specified window. Other presentations plot a negative underwater curve; its lowest point has the same magnitude with the opposite sign. Monetary and percentage declines differ in units and in comparisons across account sizes. Our illustration uses only positive values and does not address an account with negative equity. [1] [2]

Our own consequence of the definition: an omitted intraday peak or decline cannot be detected from daily closes alone. If we select a subset of observations from the same positive series, using the same window definition, its calculated maximum drawdown cannot exceed that of the full series. Temporal aggregation can nevertheless hide the worst path. Likewise, a new window with a newly initialized peak may omit a loss relative to an older peak. We therefore report the start, end, valuation interval and whether the peak is carried over from earlier history. Without these details, two correct numbers may answer different questions. [1]

A deposit raises an account’s monetary value without a trading gain; a withdrawal lowers it without a strategy loss. For time-weighted returns, GIPS separates the effect of external flows and geometrically links subperiod returns. To measure performance drawdown, we need a consistent series adjusted for those flows, not an uncritical chain of balances. We must also specify the currency, valuation of open positions, income and included costs. An asset’s price series and a portfolio’s net performance are not interchangeable. Our example below therefore excludes external flows in advance and uses values after costs. [3]

The CFA methodology distinguishes the earlier peak, subsequent trough and recovery to the peak. For clarity, we separately describe the time from peak to trough, from trough to the first return to the peak, and the full interval from peak to recovery. The deepest drawdown need not be the longest period below the peak. If the series has not recovered by the end of observation, the time to recovery is still unknown; the final date is not automatically a recovery date. Comparing only the largest percentage therefore omits a substantial part of the time path. [2]

Our own algebra for positive values shows why a decline and the subsequent required growth are not the same percentage. After a loss, recovery is calculated from a smaller base. A 25% decline from a peak of 1200 USD leaves 900 USD; recovery requires 300 USD, or growth of one third from the lower value. This arithmetic identity says neither when nor whether recovery will occur. A result in nominal currency also does not itself mean that original purchasing power has been restored. The definition of recovery must therefore match whether the series is nominal or real. [2]

Maximum refers to the observed series, not to every possible future. The CFTC notes that past performance does not predict future results. The NFA also describes limitations of hypothetical results, hindsight and imperfect representation of liquidity or slippage. A small drawdown in one selected test therefore does not guarantee a small drawdown in actual trading. Actual and simulated results must be distinguished, material assumptions disclosed, and situations outside the selected record considered. Our short fictional series merely checks the calculation. [4] [5]

Max Drawdown describes the resulting path; it is not a mechanism that stops losses by itself. In a trading plan, CME separates intended leverage, maximum trade loss, daily loss and total account exposure. Historical data do not replace those choices. A predetermined response to drawdown should specify what is measured and when further risk is reduced; even that does not turn an observed historical maximum into an execution guarantee. To assess a strategy, we therefore supplement the figure with the value path, recovery time and explicit risk-management rules, rather than only an appealing percentage. [6] [5]