Drawdown definition: what it means in practice
Drawdown is a measure of how far a performance series falls from a previously observed high. In risk discussion, the series is typically an equity curve (or another value that changes with gains and losses). A drawdown is not a prediction; it is a description of a decline that already occurred, based on how the series is defined and sampled.
A practical, checkable way to state the core mechanics is:
- Pick a time-ordered series (e.g., equity over time).
- Identify a peak value up to each point.
- Compute the percentage (or absolute amount) decline from that peak to the current value.
- The maximum such decline over a window is the maximum drawdown.
Even when the idea is simple, “advanced considerations” come from the details: which series you use, how you treat cash flows, how you measure time, and how you handle periods where the trading account value may not reflect risk the same way across platforms.
An easy model you can use to define drawdown
To make the definition precise, use an explicit and consistent calculation model. For example, define an equity-like series E(t) sampled at discrete timestamps t1, t2, …, tn.
- Peak tracking For each index i, define the running peak:
- P(i) = max{E(1), E(2), …, E(i)}
- Drawdown at each point Define drawdown (percentage form) as:
- DD(i) = (E(i) − P(i)) / P(i)
This yields DD(i) ≤ 0, with 0 meaning the series is at its peak.
- Maximum drawdown over a window Over i = k … n, define:
- MaxDD = min{DD(i)}
In words: the deepest decline from any prior peak within the window.
Advanced consideration: you must state whether you report absolute drawdown (E(i) − P(i)) or percentage drawdown (normalized by P(i)). Percentage drawdown behaves differently than absolute drawdown when the account level changes.
Advanced dependencies: assumptions that change the result
1) Choice of series (equity, balance, or “value”)
“Drawdown definition” depends on what you measure. Common candidates include equity, which reflects both realized and unrealized outcomes, and balance, which may reflect only realized results. If unrealized components are included, the drawdown can change when market prices move even if you did not close positions.
If you use account value that includes open positions, the drawdown definition is sensitive to mark-to-market conventions. If you use realized-only measures, the drawdown definition becomes less responsive to intra-trade volatility but can understate temporary risk.
2) Handling deposits and withdrawals
Cash flows can create artificial peaks and troughs. If you do not separate performance from funding events, a deposit can raise the series and reduce the apparent drawdown, while a withdrawal can lower the series and increase it.
Advanced approach: decide whether the definition is meant to describe trading performance only, or trading plus cash flow. The definition must state the assumption. If the goal is performance risk, you need a method to adjust for external cash movements so that the drawdown reflects the strategy’s effect, not the funding decision.
3) Time sampling and window definition
Drawdown values depend on when and how often you sample the series. If you sample infrequently, the recorded trough might miss the true low between timestamps, leading to an understated drawdown. If you sample very frequently, you may capture noise that is not economically meaningful.
Also, define the window clearly (full history, rolling windows, calendar months, or “since last reset”). Maximum drawdown differs across these choices.
4) Percentage normalization edge cases
Percentage drawdown divides by the peak P(i). If the series can approach or cross zero, normalization can become unstable or misleading. Even if your account value stays positive, small peaks can exaggerate percentage drawdowns relative to absolute declines.
Advanced consideration: specify whether you restrict analysis to periods where P(i) is above a threshold, or whether you accept the instability as part of the definition.
5) Margin and liquidation dynamics
In leveraged systems, account value can move quickly due to unrealized losses and margin constraints. A definition based on equity may show steep drawdowns, while the practical “risk realization” may occur through forced exits or margin calls.
Failure mode: two systems with similar equity drawdowns can have different paths to the worst point (gradual decline vs. abrupt stop-outs). Drawdown magnitude alone may not describe survivability or the time needed to recover.
6) Costs and execution assumptions
Transaction costs, financing, and execution behavior affect the equity series, and therefore the drawdown series. If you compare definitions across sources, you must know whether the data includes fees and costs consistently.
Advanced consideration: if costs are omitted or modeled differently, the drawdown definition may measure different underlying economics.
Evidence or example: how two implementations can disagree
Consider the same conceptual performance, but two different implementations:
- Implementation A reports drawdown based on an equity series that includes unrealized changes.
- Implementation B reports drawdown based only on realized gains/losses.
If a strategy experiences a large temporary loss while positions are open, Implementation A will record drawdown during that period. Implementation B may show smaller or delayed drawdown because the realized equity does not reflect the unrealized decline.
Another discrepancy comes from cash flows: if you add funds mid-period, the peak and trough relationships change. A drawdown computed without adjusting for deposits may appear less severe than the trading performance actually was.
Advanced consideration: in any example you present, state assumptions explicitly—what E(t) means, whether cash flows are excluded, the sampling frequency, and whether percentage or absolute drawdown is reported.
Limitations and risks: material failure modes
Drawdown is path-dependent and incomplete
Drawdown magnitude does not capture how long the system stays depressed or how quickly it recovers. Two strategies can share a similar maximum drawdown but differ in recovery time. Without additional measures (like time-to-recovery) drawdown definition alone can be insufficient.
Measurement can hide “risk near the edge”
If the series definition does not reflect margin constraints accurately, drawdown might underestimate the probability of forced exit. Conversely, it might overstate stress by marking unrealized losses that do not represent realized risk.
Historical drawdown does not imply future drawdown
A key limitation: past drawdowns do not establish future results. Market conditions, volatility regimes, and execution conditions can change, altering the drawdown profile.