The short answer
Maximum drawdown is the largest percentage fall from a previous peak to a later low in a series of values. At every observation, compare the value with the highest value seen so far: the smallest of those comparisons is the maximum drawdown.
The gain needed to recover from that low is a separate number. For any fall between 0% and 100% it is larger than the fall: a 25% fall needs a 33.33% rise to get back to the peak. If the series never falls below a previous peak, the maximum drawdown and the gain needed are both 0.
The formula
For observations in time order, keep a running peak:
- running peak at point t = the highest value from the first observation up to t
- drawdown at point t = value at t ÷ running peak at t − 1
- maximum drawdown = the smallest (most negative) drawdown over all points
- gain needed to recover = peak ÷ trough − 1, for the peak and trough of the maximum drawdown (0 when there is no drawdown)
Every value must be greater than zero; the maximum drawdown tool on this site accepts only positive values. A trough of 0 would be a 100% fall, from which no percentage gain can recover, so the recovery formula does not apply to it.
This site shows drawdowns as negative percentages, so −25.00% means a 25% fall. Some software shows the same fall as a positive 25%; the size is the same, only the sign convention differs.
A worked example
Four illustrative observations, in time order: 100, 120, 90, 150. This is example data, not any real security.
| Point | Value | Running peak | Drawdown |
|---|---|---|---|
| 1 | 100 | 100 | 0.00% |
| 2 | 120 | 120 | 0.00% |
| 3 | 90 | 120 | −25.00% |
| 4 | 150 | 150 | 0.00% |
The smallest drawdown is at point 3: 90 ÷ 120 − 1 = −25.00%. So the maximum drawdown is −25.00%, from the peak at point 2 (120) to the trough at point 3 (90).
To get from 90 back to 120 needs 120 ÷ 90 − 1 = 33.33%. Point 4 (150) is above the old peak, so this series has recovered.
Change the last value to 100 (100, 120, 90, 100) and the maximum drawdown is still −25.00%, but the series has not recovered: 100 is still below the peak of 120.
You can recompute both series with the maximum drawdown tool; the first one loads with its example button.
Limits and common misreadings
- It depends on how often you look. The same history measured with daily values can show a deeper maximum drawdown than with monthly values, because a monthly series can skip the lowest day.
- It describes the past, not the future. A past maximum drawdown is not a limit on future falls.
- It is not volatility. A series can move up and down a lot and still have a small maximum drawdown, or move smoothly into one long fall.
- The fall and the recovery are not symmetric. Because the recovery is measured from the lower value, a fall of x (with 0 < x < 100%) needs a rise larger than x to recover.
- Only the observations you enter are used. The result says nothing about values between them, dividends, fees or cash added or withdrawn.