Skip to main content

Drawdown & Recovery

Updated Jun 24, 2026 at 2:35pm

Research Draft High 1,214 words

A drawdown is the peak-to-trough decline in the value of an account, strategy, or asset, measured as a percentage from the highest prior equity high to a subsequent low. Maximum drawdown (MDD) is the largest such decline over a period — the single deepest "hole" the equity curve fell into. Recovery is the climb back from the trough to a new equity high. The core tension of the topic is an arithmetic asymmetry that the metric makes visceral: a loss and the gain needed to undo it are not equal, and the gap widens sharply as losses deepen. Drawdown matters because it is the dimension of risk a trader actually experiences — volatility is abstract, but watching capital fall 30% from its high is the thing that triggers capitulation, deleveraging, and abandoned strategies.

How it's calculated / formed

Drawdown at time _t_: DD_t = (Trough_t − Peak) / Peak, where Peak is the running maximum of equity up to _t_. It is always ≤ 0 (often quoted as a positive magnitude).

Maximum drawdown: the most negative value of DD_t across the whole series — the deepest peak-to-trough fall before a new high was made.

Recovery gain (the asymmetry). To return from a drawdown of size _d_ back to the prior peak requires a gain on the reduced balance of:

Required gain = 1 / (1 − d) − 1

Because the gain compounds off a smaller base than the loss subtracted from, the required recovery always exceeds the loss, and accelerates (commonly cited figures, all from the same formula):

DrawdownGain to recover
10%11.1%
20%25%
30%42.9%
50%100%
75%300%

Two related concepts: time underwater (calendar time from the prior peak until a new high is reached) and recovery time / recovery factor (net profit ÷ max drawdown).

Drawdown-based risk ratios compress this into one number:

  • Calmar ratio — CAGR ÷ MDD over a rolling 36-month window. Created by Terry W. Young, published in Futures (1991); the name is from his firm, CALifornia Managed Accounts Reports.
  • MAR ratio — CAGR ÷ MDD over the full track record since inception (associated with the Managed Accounts Reports newsletter). Differs from Calmar mainly in lookback window, which hurts cross-fund comparability when track records differ in age.
  • Ulcer Index (UI) — root-mean-square of the percentage drawdowns over a period; unlike MDD it captures both depth and duration of underwater periods. The Ulcer Performance / Martin ratio is excess return ÷ UI.

How it's used in practice

  • Sizing the worst-case, not the typical case. Position sizing and leverage are often set so that a plausible adverse run keeps MDD within a stated tolerance. Many professional and CTA programs target maximum drawdowns under ~20%; the recovery math is why — beyond ~30%, the gain required to recover grows faster than the loss and risk-of-ruin rises steeply (commonly cited practitioner heuristic).
  • Manager / strategy comparison. Calmar and MAR let an allocator rank strategies by return per unit of worst pain rather than per unit of volatility — useful because two strategies with identical Sharpe can have very different drawdown profiles.
  • Drawdown governance. Funds and prop firms set hard drawdown limits (daily/monthly/trailing) that force de-risking or a trading halt when breached. The trailing max-drawdown rule is standard at funded-trader firms.
  • Backtest validation. A backtest's MDD and its longest time-underwater are headline robustness checks: a strategy whose worst historical drawdown is near the limit of tolerance, or that spent years underwater, is fragile even if its CAGR looks good.

Adoption, debate & evidence

Drawdown is near-universal in institutional risk reporting (hedge funds, CTAs, allocators) and ubiquitous in retail/prop-firm rules. It is genuinely useful, but its single most cited weakness is that MDD is a path-dependent sample statistic, not a stable parameter. It is determined by the ordering of returns: cluster the losses and MDD deepens; spread the same losses out and it shrinks. A reported max drawdown is therefore one realized path out of many the same return process could have produced — closer to a sample maximum than to a reliable ceiling (Quantreo; PortfoliosLab). Out-of-sample drawdowns frequently exceed the worst in-sample drawdown. This is why drawdown is best treated as one axis of a multi-metric framework alongside volatility and VaR, not a standalone risk number.

On recovery, the historical record is sobering and well documented. For the S&P 500 (price index): the 1929–1932 crash fell ~82–86% and took into the 1950s to reclaim the nominal peak; the 2000–2002 and 2007–2009 declines (~49% and ~57%) each took roughly 4–7 years (and on some price-only measures over a decade) to make new highs; the 2020 COVID crash was the fastest 30%+ drawdown on record (about a month peak-to-trough) and recovered in about five months (Morgan Stanley Counterpoint Global; QuantifiedStrategies). General pattern: recovery typically takes considerably longer than the decline. (Note: total-return and inflation-adjusted recovery clocks differ materially from price-only figures — sources disagree on exact recovery durations precisely because of this.)

Strengths & limitations

Strengths: Intuitive and behaviorally honest — it measures the loss a holder must actually sit through, which is what drives real-world abandonment of sound strategies. Drawdown-based ratios (Calmar, Ulcer/Martin) penalize tail and duration risk that Sharpe largely ignores.

Limitations / failure modes:

  • Path dependence & instability (above) — the number is noisy and understates true downside potential, especially from short samples.
  • Single-point focus. Plain MDD reflects one episode and ignores the frequency and shape of all other drawdowns (UI partly fixes this).
  • Length bias. Longer track records mechanically tend to show larger MDDs, distorting MAR comparisons across funds of different ages.
  • #1 misuse: treating a historical/backtested max drawdown as a hard floor on future losses ("it never lost more than X, so X is the worst case"). It is a sample observation, not a guarantee — future drawdowns routinely exceed it.

Sources

Flagged dispute: exact S&P 500 recovery durations vary by source depending on price-only vs total-return vs inflation-adjusted basis — figures above are approximate and basis-dependent. The "Calmar = Terry Young 1991 / MAR = Managed Accounts Reports" attribution corrects a common conflation in secondary sources that credit Calmar to "Leon Rose."