What Drawdown Means and Why It Matters

Drawdown is the percentage decline from a portfolio's highest value to a later low, measured before a new high resets the reference point. It answers one question: how far below its best level did the portfolio fall, and how long did it take to get back? The number describes the past path of an investment, not its future.

Key takeaways

Drawdown is measured from the most recent peak, so a portfolio can show many drawdowns, and the "maximum drawdown" is the deepest one in the period examined.
Recovery is asymmetric: a 20% drawdown needs a 25% gain to get back to the old peak, and a 50% drawdown needs a 100% gain.
Depth is only half the picture. Daily closes of the S&P 500 price index show a 33.9% drawdown in 2020 that kept the index below its old peak for 181 days, and a shallower 25.4% drawdown in 2022 that kept it there for 746 days.
The same holdings can show different maximum drawdowns depending on data frequency, dividend treatment and the length of the record.

How drawdown is measured

Two readings are needed at every date: the portfolio's current value and the highest value it has reached so far, called the running peak. Drawdown is the gap between them, expressed as a percentage of the peak. A CFA Institute blog post by Prasad Ramani, CFA describes maximum drawdown as the accumulated loss of buying at a local high and selling at a later low, and notes that an investment can have many drawdowns but only one maximum.

Drawdown(t) = Value(t) / RunningPeak(t) - 1
Maximum drawdown = the lowest Drawdown(t) in the period

RunningPeak(t) = the highest Value on or before date t
Gain needed to recover = 1 / (1 + Drawdown) - 1

Because the running peak only moves up, drawdown is zero at every new high and negative everywhere else. Three durations are usually read alongside the depth:

Fall: the time from the peak to the trough, the lowest point before a new high.
Recovery: the time from the trough until the value first returns to the old peak.
Time under water: fall plus recovery, the whole stretch spent below the old peak.

Investor.gov, the SEC's investor education site, defines risk as the degree of uncertainty about the rate of return and the potential harm when returns differ from what the investor expected. Drawdown puts a number on one form of that harm: the size of the decline actually experienced.

A worked example, month by month

Illustrative example (hypothetical numbers, not a real portfolio). A portfolio starts at an index value of 100 and is observed at the end of each of 24 months. The table shows the months that matter; the full series is 100, 104, 108, 112, 110, 104, 98, 92, 90, 94, 99, 105, 109, 112, 115, 118, 112, 108, 111, 114, 117, 119, 121, 124 for months 0 to 23.

Month	Value	Running peak	Drawdown
3	112	112	0.0%
6	98	112	-12.5%
8	90	112	-19.6%
11	105	112	-6.2%
13	112	112	0.0%
15	118	118	0.0%
17	108	118	-8.5%
21	119	119	0.0%

The first drawdown starts at the month-3 peak of 112 and bottoms at 90 in month 8, which is 90 / 112 - 1 = -19.6%. The fall takes 5 months and the recovery to 112 takes another 5, so this episode lasts 10 months. The portfolio then needs a gain of 112 / 90 - 1 = 24.4% from the low just to match the old peak.

The second drawdown is shallower and shorter: from 118 in month 15 to 108 in month 17 (-8.5%), back above the peak in month 21. The maximum drawdown for the full 24 months is therefore -19.6%, even though the portfolio finishes at 124, 24% above where it began. In 14 of the 24 monthly readings the value sat below its running peak, which illustrates how a portfolio that ends higher can still spend most of its time under water.

Why recovery needs a bigger gain

A loss shrinks the base that the next gain acts on. After a 20% drop, 80 cents of every dollar remain, and 80 cents must grow by 25% to reach a dollar again. The required gain is 1 / (1 + drawdown) - 1, and it rises faster than the drawdown does.

Drawdown	Gain needed to recover
-10%	+11.1%
-20%	+25.0%
-30%	+42.9%
-50%	+100.0%
-75%	+300.0%

The asymmetry explains why drawdown is read differently from an average return. A shallow drawdown can be recovered by ordinary gains, while a deep one needs a much larger gain, and the time spent waiting is part of the cost. Depth alone does not set that time: in the S&P 500 episodes below, the deeper 2020 decline was recovered far sooner than the shallower 2022 one. Whether the money is needed during that wait, for example for regular withdrawals, changes the outcome; that interaction is usually discussed as sequence-of-returns risk.

Real US stock drawdowns

The Federal Reserve Bank of St. Louis publishes daily closes of the S&P 500 on FRED, sourced from S&P Dow Jones Indices. It is a price index without dividends, and FRED carries only the latest 10 years of history. Computing running-peak drawdowns on that series from October 17, 2016 through October 16, 2026 gives these episodes deeper than 15%. Days are calendar days, and "recovery" means the first close at or above the old peak close.

Peak close	Trough close	Drawdown	Peak to trough	Trough to recovery	Gain needed
Sep 20, 2018	Dec 24, 2018	-19.8%	95 days	120 days	+24.7%
Feb 19, 2020	Mar 23, 2020	-33.9%	33 days	148 days	+51.3%
Jan 3, 2022	Oct 12, 2022	-25.4%	282 days	464 days	+34.1%
Feb 19, 2025	Apr 8, 2025	-18.9%	48 days	80 days	+23.3%

Fall speed and recovery speed varied widely. The 2020 drawdown fell in 33 days and was fully recovered 181 days after the peak. The 2022 drawdown fell more slowly and left the index below its January 2022 close for 746 days. Both episodes, like the other two, are visible only with hindsight: a trough can be identified only after a new high has been reached.

Earlier history is deeper. Yardeni Research's tables of S&P 500 bull and bear markets, based on Standard & Poor's data, list closes of 1,565.15 on October 9, 2007 and 676.53 on March 9, 2009. That is a 56.8% drawdown over 517 days, which required a gain of about 131% to recover. The same tables date the next record high to March 28, 2013, 1,997 days after the 2007 peak. These are price levels, so dividends are excluded here as well.

Where drawdown figures differ

Two sources can report different "max drawdown" numbers for the same holdings. Four conventions explain most of the gap.

Data frequency

A drawdown can only be measured at the dates observed. Sampling the hypothetical series above every third month, starting at month 0, shows a maximum of -16.1%; starting from month 1 or month 2 gives -16.4% or -16.7%. Monthly readings gave -19.6%, so every quarterly version missed the peak, the trough or both. A sample of fewer dates can never show a deeper drawdown than the full series. A fund page based on monthly statements and a dashboard based on daily prices will rarely agree.

Price versus total return

A price series leaves out dividends, as the FRED series above does. A total-return series adds them back as if reinvested. Because dividends can only add to value, a total-return drawdown is never deeper than the price drawdown over the same dates, and it ends no later; the difference is roughly the dividends paid during the decline, so it is small for a fall lasting weeks. Fees pull in the other direction, so check whether returns are net of fund expenses.

Length of the record

Maximum drawdown tends to grow with the length of the record. A peer-reviewed paper by Magdon-Ismail, Atiya, Pratap and Abu-Mostafa in the Journal of Applied Probability (2004) studies the maximum drawdown of a Brownian motion with drift, a simplified mathematical model of a random price path. Its abstract states that, as the horizon lengthens, the expected maximum drawdown grows logarithmically when the drift is positive, with the square root of time when it is zero, and linearly when it is negative. In plain terms, a 20-year record and a 3-year record are not comparable on maximum drawdown alone.

The starting point of the chart

Some charts reset the peak at the start of the selected window; others use all available history. A window that opens just after a high will understate the earlier decline, and one that opens at a high will count the whole fall.

What drawdown does not tell you

Drawdown is a description of a past path, and it has limits.

It is not a forecast. The deepest past decline is not a floor. FINRA's guide to investment risk states that all investments carry some degree of risk and that stocks, bonds, mutual funds and ETFs can lose value.
It is a different measure from volatility. Volatility captures how widely returns swing, while drawdown captures the cumulative fall from a high; two portfolios can rank differently on each.
It does not say why. The figure gives depth and timing, not the causes, so exposures still need to be examined separately.
It depends on pricing. Assets that are valued only occasionally are observed at fewer dates, and those values may lag the market, so the reported drawdown can understate the decline.
It does not include behavior. A drawdown that was recovered on paper still costs money for someone who sold during it.

Labels such as correction and bear market are conventions for drawdowns of particular depth, and a fund's prospectus, which Investor.gov describes as covering objectives, risks, past performance and expenses, shows past performance in its own format, so the depth and duration of declines may need to be recalculated from the return series.

Frequently asked questions

Is maximum drawdown the same as the worst year's loss?

No. A calendar-year return compares two year-end values, while drawdown tracks every date in between. In the S&P 500 price data above, 2020 ended with a 16.3% gain even though the index had been 33.9% below its peak in March, and 2022 ended with a 19.4% loss against a 25.4% maximum drawdown.

What is the Calmar ratio, and how does it use drawdown?

The CFA Institute blog cited above defines it as the compounded annualized growth rate divided by the maximum drawdown over the same period. It adds that, unlike the Sharpe ratio, the Calmar ratio cannot be scaled across periods, so portfolios must be compared over identical windows. For the hypothetical series, 11.9% annualized growth over 23 months divided by the 19.6% maximum drawdown gives about 0.6.

When does a drawdown end?

The trough marks the bottom, but the drawdown as a whole ends when the value returns to the previous peak. Until then, any further decline below the trough deepens the same drawdown rather than starting a new one. A new drawdown begins only after a new high has been set.

Can a drawdown be worse than minus 100%?

For an unleveraged, long-only holding, no, because a value cannot fall below zero, so the maximum is a 100% loss. Positions bought with borrowed money can lose more than the amount put in: Investor.gov's bulletin on margin accounts shows a stock bought for $50 on margin that falls to $15 costing the investor more than 100% of their money. A drawdown measured on the capital invested can therefore exceed 100%. Most published figures are for unleveraged funds and indexes.

Do index figures include dividends?

It depends on the series. The FRED S&P 500 data used here is a price index and, according to FRED's own series notes, does not contain dividends, so a total-return version of the same index would show drawdowns that are equal or slightly shallower. Always check which version a chart uses before comparing two numbers.

A checklist for reading a chart

Confirm the date range and whether the peak is reset at the start of the window.
Check the data frequency (daily, monthly or quarterly) and the pricing source.
Check whether returns include dividends and are net of fees.
Record the depth, the days or months of fall, and the time to recover for the worst episode, not just the percentage.
Calculate the gain needed from the trough with 1 / (1 + drawdown) - 1 to see what recovery would require.
Compare two portfolios only over identical windows and conventions.