Drawdown Recovery Analysis: Underwater Curves and the Arithmetic of Losses

Drawdown recovery analysis measures how long a portfolio stays below its previous high, not only how far it fell. Depth answers how bad it got. Recovery answers how long it stayed bad, and for most real investors the second question is the one that decides whether they are still invested when the recovery arrives.

The tool that shows both at once is the underwater curve, and the reason the analysis matters is a piece of arithmetic that is easy to state and hard to internalise: losses and the gains needed to reverse them are not symmetric.

The arithmetic of getting back

Start with the mechanics, because everything else follows from them.

A drawdown is measured from a running peak. If a portfolio reaches a high and then falls, the drawdown on any later date is the percentage gap between the current value and that highest previous value. It stays measured against that peak until a new peak is set, at which point the drawdown resets to zero.

The recovery required is where the asymmetry appears. A gain is calculated on the reduced base, so it must be larger than the loss to get back to level.

Fall from peak	Gain required to recover
10 percent	About 11 percent
20 percent	25 percent
33 percent	About 50 percent
50 percent	100 percent
66 percent	About 200 percent
80 percent	400 percent

These are arithmetic identities, not estimates, and they hold for any portfolio in any market. The relationship is convex: the required gain rises far faster than the loss. That is the reason risk work concentrates so heavily on avoiding the deep left tail rather than on optimising the middle of the distribution. A shallow drawdown is an inconvenience. A deep one changes the mathematics of the rest of the investment life.

The second half of the story is time. Recovering one hundred percent is not a single event, it is a period. At a compounding rate that any long term investor would consider good, doubling takes years. The deeper the hole, the longer the climb, and the length of that climb is the thing an investor actually lives through.

The underwater curve

The underwater curve is the standard visual. On every date it plots the portfolio’s percentage below its running peak. The line sits at zero whenever a new high is being made, and drops below zero in between.

What to read from it:

Depth of the troughs. The lowest point is the maximum drawdown, which is the single most quoted risk statistic and the one this chart puts in context.

Width of the troughs. How long each episode lasted from peak to recovery. A wide, shallow trough and a narrow, deep one can have very different psychological weight even where the depth statistic is similar.

Proportion of time underwater. How much of the whole record was spent below a previous high. A strategy that spends most of its life underwater, making new highs only briefly, is a very different proposition from one that grinds upward with short interruptions, even where the end points match.

Frequency. How often the portfolio goes underwater at all. Frequent shallow dips are a different risk profile from rare deep ones.

Two decompositions are worth naming. Time to trough is how long the fall took, from peak to the lowest point. Time to recovery, sometimes called the recovery period, is from the trough back to the old peak. The total, peak to peak, is the underwater period. A fast crash with a fast rebound and a slow grind down with a slow grind back can produce the same maximum drawdown while being completely different experiences.

How to read it well

Read depth and duration together, never separately. A maximum drawdown quoted alone is half the information. Pair it with the longest underwater period and the picture becomes usable. This is the same reason the Calmar ratio, which divides return by maximum drawdown, is an improvement on raw return and still incomplete: it uses depth and ignores duration.

Check whether the recovery actually happened. An underwater curve that ends below zero means the portfolio has not recovered yet. Any statistic computed as though the episode were complete, including the recovery period, is truncated and will understate the true figure. This is a routine error in performance reporting.

Look at the worst few, not just the worst one. Maximum drawdown is a single observation and therefore a noisy statistic. The distribution of the five or ten largest drawdowns says much more about what to expect, and it depends far less on one unlucky window.

Mind the measurement frequency. Drawdown computed on monthly data will understate what daily data shows, because intra month troughs are invisible. Two records are not comparable unless they were measured at the same frequency, and a monthly figure quoted next to a daily one flatters the monthly one.

Mind the window. Drawdown statistics depend heavily on the period covered. A record that happens to start after a major decline and end before the next one will show a benign underwater curve that says more about the window than the strategy. Longer windows including at least one full cycle are the minimum for a fair reading.

Distinguish price falls from permanent impairment. A portfolio can be underwater because the market repriced everything, or because the underlying businesses deteriorated. The chart looks identical. Only fundamental work separates the two, which is why drawdown analysis belongs next to research on the holdings rather than instead of it.

Use it to plan in advance. The practical value of studying past underwater periods is that it sets an expectation before the next one starts. A committee that has looked at how long previous recoveries took is in a different position from one encountering the question for the first time in the middle of a decline. That is the connection to stress testing a portfolio, where the shocks are designed rather than observed.

What it does not tell you

It is entirely backward looking. Every drawdown statistic describes what already happened in one particular sample. The worst drawdown in a record is the worst so far, and there is no principle that makes it a ceiling. Treating a historical maximum as a limit on future losses is the most common misuse of the measure.

It says nothing about why. The curve records that value fell below a peak. It does not distinguish a broad market decline from a sector rotation from a genuine deterioration in the holdings, and the correct response to each is different.

Cash flows distort it. For a portfolio with contributions and withdrawals, the value series mixes market performance with money movement, and a drawdown measured on that series is not a clean statement about performance. Comparing the two requires the distinction covered in CAGR vs XIRR vs absolute returns.

Recovery to a nominal peak is not full recovery. Returning to the previous value ignores inflation and the opportunity cost of the years spent getting there. A portfolio back at its old level after a long underwater period has recovered in the chart and not in purchasing power.

It does not capture how the drawdown felt or how it was funded. Whether an investor could hold through it depends on liabilities, redemption pressure, mandate limits and temperament, none of which appear on the curve. The largest permanent losses usually come from decisions made during a drawdown rather than from the drawdown itself.

Small differences in depth are noise. Maximum drawdown is a single extreme observation and moves a lot with the sample window and measurement frequency. Ranking strategies by small gaps in maximum drawdown reads precision into a statistic that does not support it.

A backtested underwater curve is optimistic by default. Simulated drawdowns typically exclude costs, slippage and the difficulty of trading during a decline, and they assume perfect discipline throughout. The realised curve is generally deeper and longer, for the reasons set out in common backtesting mistakes.

The lasting takeaway is the convexity. Because recovery requirements grow faster than losses, the difference between a moderate and a severe drawdown is not a difference of degree but of kind. That is why serious risk work spends so much of its attention on the deep left tail, and why the length of the road back deserves as much space in a review as the depth of the fall.

Related reading

• Portfolio and Backtest Metrics, Explained: the hub guide to the metrics and methods behind portfolio analysis.
• What Is Maximum Drawdown?: the largest peak to trough fall and why it matters more than volatility.
• Stress Testing a Portfolio: designing severe but conceivable shocks before they arrive.
• What Is the Calmar Ratio?: return measured against the worst drawdown, and what it leaves out.
• Monte Carlo Simulation in Investing: simulating a distribution of drawdowns rather than reading one realised path.

This article is educational. Altys Labs is not a registered research analyst or investment adviser, and nothing here is investment advice or a recommendation to buy, sell, or hold any security.