Lesson 8.7 · about 8 minutes
Drawdown maths
- Hypothetical
- Long history
Learn to
- Convert a fall into the gain needed to get back
- Estimate recovery time at a steady yearly gain
- See why an average yearly return can overstate what a pot grew by
Ideas in this lesson
- Drawdown
- Gain needed to recover
- Recovery time
- Arithmetic and geometric average
- Volatility drag
A fall of 50% needs a gain of 100% to get back, because the gain applies to a smaller base. The general rule: a fall of f needs a gain of f ÷ (1 − f). A fall of 10% needs about 11.1%, 33% needs about 49%, and 70% needs about 233%. This is why the size of a fall matters more than how often falls happen. The next lessons use the word drawdown for the fall from a high.
Time matters as well. At a steady 8% a year, a pot that fell 50% needs about 9 years just to return to its old level (the natural log of 2 divided by the natural log of 1.08). That is recovery time, the stretch most people actually feel.
Volatility drag: a pot that gains 50% in one year and loses 50% in the next has an arithmetic average of 0%, yet ends 25% lower (1.5 × 0.5 = 0.75). The steady yearly rate that gives the same end value, the geometric average, is about −13.4%. The more a pot swings, the further the geometric average falls below the arithmetic one.
Try it · hypothetical fall, then real falls in past data
Slide the size of a fall, set a steady yearly gain, then read how long real falls in the data took to recover.
The gain a fall needs
Slide the loss. The gain needed to get back is always bigger, and the gap explodes as the loss gets deeper.
To turn ₹50,000 back into ₹1,00,000 you need +100%. A 50% fall followed by a 50% rise leaves you at ₹75,000.
This is the loss slider from Reading backtest results (module 2), reused as it is.
Recovery time at a steady yearly gain
Gain needed to get back
66.7%
Years to get back
6.6
80 months
₹10 lakh falls to
₹6,00,000
Volatility drag: up by a share, then down by the same share
Arithmetic average of the two years
0.0%
Steady yearly rate that gives the same end value
−13.4%
₹1 lakh ends at
₹75,000
Loading the long-history data…
The last panel lists every fall of 10% or more in the Nifty 50 price index in our data, with how long the fall took and how long the climb back took. The first one starts on the first day of our data, so its real fall from an earlier high is at least what is shown. The index is a price index, so dividends would have shortened these recoveries a little. It is past data, not forecast.
For the fall of one investor's portfolio, the same rule applies to the portfolio's own value, whatever its mix.
Check yourself
0 of 3 right
Answer every question correctly to complete this lesson.
Sources
Last reviewed 30 Oct 2026 by the author. Independent fact review is pending before public launch. Facts are as of the dates shown. Education only; not investment advice.