Drawdown

Drawdown is computed from a series of portfolio values, not from returns in isolation. At each observation date, record the highest value reached up to and including that date, the running peak. The drawdown at that date equals the current value divided by the running peak, minus one, expressed as a negative percentage. When the portfolio sets a new high, the running peak resets and the drawdown returns to zero. The result is a continuous series that is zero at every new high and negative in between.

Maximum drawdown is simply the most negative value in that series over the measurement window. Because the running peak depends on the order in which returns arrive, drawdown is path-dependent: the same set of monthly returns reshuffled produces a different maximum drawdown, whereas the mean and standard deviation of those returns are unchanged.

The gain required to recover from a decline is always larger in percentage terms than the decline itself, because the rebound is applied to a smaller base. Formally, the required return equals one divided by the quantity one minus the drawdown fraction, minus one. A 20% decline requires 25% to break even, a 33% decline requires 50%, and losses beyond that escalate sharply. This convexity is why capital preservation dominates downside risk management for portfolios with finite horizons.

Depth alone is therefore incomplete. Practitioners also report the drawdown duration, or time under water: the interval from the prior peak through the trough to the date the peak is finally exceeded. Two funds can share an identical maximum drawdown while one recovers in months and the other remains below its high for years, which matters greatly for investors facing withdrawals or liability payments during the recovery.

Drawdown is embedded in the commercial architecture of asset management. Hedge fund high-water mark provisions withhold incentive fees until a fund exceeds its previous peak, so an active drawdown directly suppresses manager compensation. Managed futures and CTA programmes commonly publish maximum drawdown alongside return and volatility, and investment policy statements may specify drawdown tolerances or trigger levels that force de-risking.

Risk-adjusted measures build on it. The Calmar ratio divides annualised return by maximum drawdown, ranking managers by return earned per unit of worst realised loss, and related statistics such as the Ulcer Index average the depth of drawdowns over time rather than taking only the extreme. Leveraged and margined positions add a mechanical dimension: a sufficiently deep drawdown can trigger margin calls or redemption gates that force liquidation at the trough, converting a paper loss into a permanent one.

A candidate should be able to take a short series of portfolio values or returns, track the running peak, compute the drawdown at each date, and identify the maximum. Equally important is positioning drawdown against other risk metrics: standard deviation penalises upside and downside deviation symmetrically and ignores sequencing; downside deviation and semivariance isolate returns below a threshold; Value at Risk and expected shortfall attach probabilities to a horizon. Drawdown alone attaches no probability to anything.

Its principal weaknesses are sample dependence and observation frequency. Maximum drawdown is a single realised extreme, so it is unstable across periods and tends to grow simply because a longer history offers more chances for a bad run. Monthly valuations also understate true peak-to-trough moves that daily data would capture, which flatters illiquid or infrequently priced strategies.