A drawdown is the percentage fall in an investment's value from a peak to a later low, and it stays open until the old peak is regained. It answers a question that average returns and volatility skip: how far down was someone who bought at the top at the worst moment? Maximum drawdown is the deepest such fall over a chosen period. Because the climb back starts from a smaller base, every recovery needs a bigger percentage gain than the fall that caused it. The stocks hub covers the index funds these figures most often describe.
How do you calculate a drawdown?
Divide the low by the highest value that came before it and subtract 1; the answer is the drawdown, usually quoted as a negative percentage or as a fall.
The highest earlier value is a running peak. It moves up each time the investment sets a new high, so a drawdown starts at a peak and ends only when that peak is equaled or beaten; a rally that falls short leaves it open. Published figures usually use daily or month-end closing prices, which means falls within a trading day are ignored. Three lengths describe each episode: peak to low, low back to the peak, and the whole stretch below the old high, sometimes called time under water.
Worked example: The S&P 500's 2022 drawdown, step by step
Closing levels from FRED. The running peak was the record close of 4,796.56 on January 3, 2022.
| Date | S&P 500 close | Drawdown from 4,796.56 |
|---|---|---|
| January 3, 2022 | 4,796.56 | 0%, the peak |
| June 16, 2022 | 3,666.77 | −23.6% |
| August 16, 2022 | 4,305.20 | −10.2% |
| October 12, 2022 | 3,577.03 | −25.4%, the low |
| January 19, 2024 | 4,839.81 | First close above the peak |
- Maximum drawdown: 3,577.03 ÷ 4,796.56 − 1 = −25.4%
- Gain needed from the low: 4,796.56 ÷ 3,577.03 − 1 = 34.1%
- Time: 282 days from peak to low, then 464 days back to a record, 746 days in all
The August rally lifted the index 17.4% from its June low, yet the drawdown stayed open because January's peak was untouched, and October brought a deeper low. The episode closed only when a daily close beat 4,796.56, which is why the recovery date sits more than 15 months after the bottom.
Why does a 50% drawdown need a 100% gain?
The recovery starts from the lower value, so a 50% fall that halves $10,000 to $5,000 needs the $5,000 to double, a 100% gain, to get back.
Write the drawdown as a positive fraction, so a 25% fall gives 1 ÷ 0.75 − 1, or 33.3%. The table runs the same sum for a range of falls, including some real ones.
| Drawdown | Left of $10,000 | Gain needed to get back |
|---|---|---|
| 10% | $9,000 | 11.1% |
| 20% | $8,000 | 25.0% |
| 25.4%, S&P 500 in 2022 | $7,457 | 34.1% |
| 50% | $5,000 | 100% |
| 56.8%, S&P 500 in 2007–09 | $4,322 | 131.3% |
| 76.7%, bitcoin in 2021–22 | $2,334 | 328.5% |
| 84.9%, bitcoin in 2013–15 | $1,514 | 560.4% |
| 90% | $1,000 | 900% |
The gain needed climbs far faster than the loss. Moving from a 10% drawdown to a 50% one multiplies the fall by five and the required gain by nine, and past 50% the numbers run away. Time stretches the damage too. Doubling at a steady CAGR of 7% takes about 10.2 years, and at 10% about 7.3; the compound interest calculator reruns the sum at other rates. Anyone still adding money during a fall gets back to even sooner, because new contributions buy at lower prices. The bear markets and corrections guide lists every S&P 500 bear market since 1946 and how long each recovery took.
What is maximum drawdown?
Maximum drawdown is the largest peak-to-low fall inside a chosen window, the worst loss a buyer at the top would have suffered before any recovery.
The window decides the answer. In the year to October 28, 2026, the bitcoin vs S&P 500 study found a maximum drawdown of 48.2% for bitcoin and 9.1% for the S&P 500, counting only prices inside that year, while bitcoin's full Bitstamp record gives 84.9%. A 10-year figure calculated today starts in late 2016 and so misses 2008, and a fund launched since then has never shown how it behaves in a crash of that size. Compare maximum drawdowns only over the same dates and sampling, and read each alongside its recovery time.
US rules give the measure legal weight in one corner of the market. Commodity pools must show their worst peak-to-valley draw-down for the latest five calendar years and the year to date in their disclosure documents, and the CFTC defines it on month-end net asset values, counted until an earlier month-end value is equaled or exceeded.
Selected drawdowns, each measured from a record to the lowest close before that record was regained:
| Series | Peak | Low | Drawdown | Back at the old peak |
|---|---|---|---|---|
| S&P 500, price only | Sep 7, 1929 | Jun 1, 1932 | 86.2% | Sep 22, 1954 |
| US market, dividends reinvested | Sep 3, 1929 | Jul 8, 1932 | 84.1% | Feb 6, 1945 |
| S&P 500, price only | Oct 9, 2007 | Mar 9, 2009 | 56.8% | Mar 28, 2013 |
| US market, dividends reinvested | Oct 9, 2007 | Mar 9, 2009 | 54.6% | Feb 28, 2012 |
| Bitcoin | Dec 4, 2013 | Jan 14, 2015 | 84.9% | Feb 23, 2017 |
| Bitcoin | Dec 16, 2017 | Dec 15, 2018 | 83.4% | Nov 30, 2020 |
| Bitcoin | Nov 8, 2021 | Nov 21, 2022 | 76.7% | Mar 4, 2024 |
| Bitcoin | Oct 6, 2025 | Jun 30, 2026, so far | 53.1% so far | Not yet |
The S&P 500 rows come from Yardeni Research's tables of S&P data, the dividend rows from Kenneth French's daily returns for the whole US market, and the bitcoin rows from Bitstamp daily closes. Stocks' worst fall was about as deep as bitcoin's but lasted far longer: more than 15 years from peak to recovery with dividends and 25 on prices alone, against just over three years for bitcoin's 2013 to 2015 crash. Dividends also shortened 2007 to 2009, putting the whole US market back at its old high 13 months before the price-only S&P 500 regained 1,565.15. At the November 1, 2026 close of $78,233, bitcoin still needed a 59.4% rise to get back to $124,728; the bitcoin drawdowns history dates all six of its falls of 50% or more from a record since 2013.
How is drawdown different from volatility, a loss and a bear market?
Volatility measures typical swings in both directions, a loss is measured from your purchase price, and a bear market is a label for a drawdown of at least 20%.
| Term | Measured from | What it tells you |
|---|---|---|
| Drawdown | The highest value before the low | How far the investment fell from its best point |
| Maximum drawdown | The highest peak in a chosen window | The deepest such fall within that window |
| Loss on a holding | The price you paid | What you are down, which depends on when you bought |
| Standard deviation | The average return | How widely returns swing, up as well as down |
| Correction | A recent high | A fall of 10% or more, in Yardeni's tables |
| Bear market | A recent high | A fall of 20% or more over at least two months, per Investor.gov |
| Pension drawdown, UK | Not a market measure | Taking income from a pension pot that stays invested |
A drawdown describes the investment, not any one investor. Someone who bought the S&P 500 at its March 2009 low never had a loss from the 2007 to 2009 fall, while someone who bought in October 2007 sat through all 56.8% of it. Bear-market tables also cut episodes differently: they restart the count after a rally of 20% or more, so Yardeni's lists split 1929 to 1932 into a 44.7% fall and an 83.0% fall, whereas measured from the September 1929 peak it was one drawdown of 86.2%.
Standard deviation can understate drawdown risk because it scores each period separately; a string of moderate monthly losses compounds into a deep fall without any single month looking extreme. The Sharpe ratio ignores the order of returns and so the drawdowns they create, which means two funds with the same ratio can have very different worst falls. In UK pensions, GOV.UK uses flexi-access drawdown for taking an adjustable income from a pot that stays invested, a separate idea, though a market drawdown early in retirement makes those withdrawals bite harder. How deep a fall a portfolio can sit through is the question behind asset allocation by age and how much crypto to hold.