Max drawdown: every S&P 500 bear market since 1929

The S&P 500's deepest decline from a closing high was −86.2%, from September 16, 1929 to June 1, 1932, and the index needed until September 1954 to close above that 1929 peak again. Since 1950 it has had 11 bear markets (closing declines of 20% or more from a record), and the median one took 24.5 months from the old high to a new one.

Those two numbers describe the risk measure most investors actually feel: maximum drawdown, the largest drop from a peak to a later low before a new peak. Volatility tells you how bumpy a ride is; drawdown tells you how far down the hole goes and how long you stay in it. This guide lists every S&P 500 decline of roughly 20% or more since 1929 from daily closing prices, shows the arithmetic of recovering from a loss, and explains how to measure drawdown on your own portfolio without deposits distorting the result. Data as of October 2026.

Every S&P 500 bear market since 1929

The table uses daily closing values of the S&P 500 price index (dividends not included). A drawdown starts at a record close and ends when the index closes above that record again; the decline is measured to the lowest close in between. FINRA describes a correction as a drop of at least 10% and a bear market as a decline of about 20% or more in a broad index, so the table also shows the four near misses that stopped just short of 20%.

Peak close	Trough close	Decline	Days peak → trough	New high	Months peak → new high
Sep 16, 1929	Jun 1, 1932	−86.2%	989	Sep 22, 1954	300.2
Aug 3, 1956	Oct 22, 1957	−21.5%	445	Sep 24, 1958	25.7
Dec 12, 1961	Jun 26, 1962	−28.0%	196	Sep 3, 1963	20.7
Feb 9, 1966	Oct 7, 1966	−22.2%	240	May 4, 1967	14.8
Nov 29, 1968	May 26, 1970	−36.1%	543	Mar 6, 1972	39.2
Jan 11, 1973	Oct 3, 1974	−48.2%	630	Jul 17, 1980	90.1
Nov 28, 1980	Aug 12, 1982	−27.1%	622	Nov 3, 1982	23.2
Aug 25, 1987	Dec 4, 1987	−33.5%	101	Jul 26, 1989	23.0
Mar 24, 2000	Oct 9, 2002	−49.1%	929	May 30, 2007	86.2
Oct 9, 2007	Mar 9, 2009	−56.8%	517	Mar 28, 2013	65.6
Feb 19, 2020	Mar 23, 2020	−33.9%	33	Aug 18, 2020	5.9
Jan 3, 2022	Oct 12, 2022	−25.4%	282	Jan 19, 2024	24.5
Near misses: Jul 16, 1990	Oct 11, 1990	−19.9%	87	Feb 13, 1991	7.0
Jul 17, 1998	Aug 31, 1998	−19.3%	45	Nov 23, 1998	4.2
Sep 20, 2018	Dec 24, 2018	−19.8%	95	Apr 23, 2019	7.1
Feb 19, 2025	Apr 8, 2025	−18.9%	48	Jun 27, 2025	4.2

Source: our calculation from daily S&P 500 closes (S&P Dow Jones Indices data; 1927–2016 via Yahoo Finance ^GSPC, 2016–2026 cross-checked against FRED series SP500). Months = calendar days ÷ 30.44.

Two features stand out. First, the depth and the length of a drawdown are only loosely related: the 2020 crash was deeper than 2022 (−33.9% vs −25.4%) yet recovered four times faster. Second, the long recoveries all followed declines that were both deep and slow — 1929–32, 1973–74, 2000–02 and 2007–09 each took more than 500 days just to reach the bottom.

1929–32−86.2%

2007–09−56.8%

2000–02−49.1%

1973–74−48.2%

1968–70−36.1%

2020−33.9%

1987−33.5%

1961–62−28.0%

1980–82−27.1%

2022−25.4%

1966−22.2%

1956–57−21.5%

The numbers worth quoting

The day-count statistics are worth reading twice. From January 1950 to mid-October 2026, the S&P 500 closed at a new record on 7.1% of trading days (about one day in fourteen), closed within 5% of its record on 44.5% of days, and spent 20.7% of all trading days at least 20% below its last high. In other words, for most of history an index investor has been sitting in some drawdown, and for about a fifth of the time in a deep one. That is normal, not a sign that something broke.

What is maximum drawdown and how do you calculate it?

Maximum drawdown (MDD) is the largest percentage fall from a running peak to a subsequent trough, over a chosen period:

drawdown on day t = value(t) ÷ highest value up to t − 1

max drawdown = the most negative drawdown in the period

Three choices change the result, so state them whenever you quote a drawdown:

1. Frequency. Daily closes catch the true bottom; month-end or year-end values smooth it away (see the section on annual data below).
2. Price or total return. A price index ignores dividends, so it shows deeper and longer drawdowns than an investor who reinvested dividends experienced.
3. Whose value. For a portfolio with deposits and withdrawals, measure the drawdown on a unit value or time-weighted return series, not on the account balance.

1. 1

   Build a value series

   Daily or at least month-end values of the portfolio, or better, of a time-weighted index that starts at 100

2. 2

   Track the running peak

   For each date, the highest value so far

3. 3

   Compute the drawdown

   Value ÷ running peak − 1; it is 0 on record days and negative otherwise

4. 4

   Take the minimum

   The most negative drawdown is the max drawdown; note its peak and trough dates

5. 5

   Measure the recovery

   Count the time from the peak until the series first closes above it again

Why losses need bigger gains to recover

A fall of x% needs a gain of x ÷ (1 − x) to get back to even, because the rebound starts from a smaller base. The gap grows quickly as losses deepen (try your own numbers in the loss recovery calculator):

Drawdown	Gain needed to break even	Years at 7% a year	Years at 10% a year
−10%	+11.1%	1.6	1.1
−20%	+25.0%	3.3	2.3
−25%	+33.3%	4.3	3.0
−33.9% (2020)	+51.3%	6.1	4.3
−49.1% (2000–02)	+96.5%	10.0	7.1
−56.8% (2007–09)	+131.5%	12.4	8.8
−86.2% (1929–32)	+624.6%	29.3	20.8

Years to recover = ln(1 ÷ (1 − loss)) ÷ ln(1 + annual return). The 7% and 10% rates are illustrative steady returns, not forecasts.

The table also explains why rebalancing and position size matter more than the precise entry price for many investors: a portfolio that never falls 50% never needs to double. (How often to bring a mix back to target is its own question — see how often to rebalance, with a 98-year 60/40 backtest.)

Annual returns hide most of the drawdown

Most published return tables are calendar-year numbers. They are useful for averages but poor at showing drawdowns, because a crash and its rebound often fall inside one year. Using annual S&P 500 total returns from 1928 to 2025 compiled by Aswath Damodaran (NYU Stern):

• 2020: the index fell 33.9% from February 19 to March 23, yet the calendar year ended +18.0% with dividends. On an annual chart, 2020 shows no drawdown at all.
• 2008–09: the year-end series shows a −36.6% drawdown (2008); the daily price series shows −56.8%, because the low came in March 2009 and the rebound before December 2009 hid it.
• Overall: on year-end total-return data the worst drawdown since 1928 is −64.8% (1929–32) — still enormous, but 21 points shallower than the −86.2% daily price figure.

The line below shows the year-end "underwater" curve of the S&P 500 total return: 0 in years that ended at a new high, negative otherwise.

	%
1995	0
1996	0
1997	0
1998	0
1999	0
2000	-9
2001	-19.8
2002	-37.4
2003	-19.7
2004	-11.1
2005	-6.8
2006	0
2007	0
2008	-36.6
2009	-20.1
2010	-8.3
2011	-6.3
2012	0
2013	0
2014	0
2015	0
2016	0
2017	0
2018	-4.2
2019	0
2020	0
2021	0
2022	-18
2023	0
2024	0
2025	0

Dividends shorten recoveries. On the price index, the 2000–02 bear market ended only on May 30, 2007. On the total-return data, $1 invested at the end of 1999 was worth $0.626 at the end of 2002, $0.932 at the end of 2005 and $1.078 at the end of 2006 — so a dividend-reinvesting investor was back above water at least five months earlier than the price chart suggests. That gap is our own derived comparison and is why you should compare your portfolio with a total-return benchmark.

Drawdown of a 60/40 portfolio vs 100% stocks

Drawdown is also the clearest way to see what diversification buys. We combined the same annual data into a portfolio of 60% S&P 500 and 40% 10-year Treasury bonds, rebalanced every January:

Year-end drawdown episode	100% S&P 500	60/40, rebalanced yearly	100% 10-year Treasuries
1929–32 (worst at end of 1932)	−64.8%	−40.1%	0.0%
1973–74	−36.5%	−20.8%	0.0%
2000–02	−37.4%	−11.7%	0.0%
2007–08	−36.6%	−13.9%	0.0%
2021–22	−18.0%	−18.0%	−21.5%
Annualized return 1928–2025	10.02%	8.34%	4.53%
Calmar-style ratio (return ÷ max drawdown)	0.15	0.21	0.21

Our calculation from Damodaran's annual return series. Year-end data understate intra-year lows.

The 60/40 portfolio gave up 1.68 percentage points of annual return but cut the worst drawdown by roughly a third and, in 2000–02 and 2008, by about two-thirds. 2022 is the exception that every 60/40 investor remembers: stocks and bonds fell together, and the 10-year Treasury series had its own worst drawdown (−21.5% across 2021–22). The SEC's guide to asset allocation describes rebalancing as "bringing your portfolio back to your original asset allocation mix" — the drawdown table shows the trade-off that mix represents.

The Calmar ratio used in the last row divides annualized return by the absolute value of the maximum drawdown: 10.02 ÷ 64.8 = 0.15 for stocks and 8.34 ÷ 40.1 = 0.21 for 60/40. Professionals normally compute it over 36 months of monthly data; here it is a long-run illustration.

How to measure drawdown on your own portfolio

The trap is measuring the account balance. Deposits raise the balance and hide losses; withdrawals deepen them. The correct series is a time-weighted unit value, the same one used for time-weighted return (TWR).

In practice: export your full transaction history, rebuild daily values, chain-link daily returns into an index that starts at 100, and run the five steps above on that index. If your accounts are spread across several brokers, combine them first — a drawdown on one account is not the drawdown of your wealth. GetHossa does the time-weighted calculation for you after you import your broker history, and its portfolio vs S&P 500 comparison shows your result over the same dates as the index, so you can see whether your worst stretch was deeper or shallower than the market's.

Common mistakes

A further edge case: a drawdown can still be "open". On October 21, 2026 the S&P 500 closed at 7,757.17, 0.8% below its October 6, 2026 record close of 7,818.93 (FRED) — a routine dip that would only enter the table if it deepened to 20%.

Key takeaways

• Max drawdown = largest fall from a running peak to a later trough; always state the frequency and whether dividends are included.
• The S&P 500's worst was −86.2% (1929–32). Since 1950 there were 11 bear markets; the median took 24.5 months from peak to new high, the longest 90 months (1973–80).
• A −50% loss needs +100% to recover; at 10% a year that is about 7.3 years.
• Annual data hide crashes: 2020 fell 33.9% but ended the year up 18.0%.
• Measure your own drawdown on a time-weighted series, not on the account balance — deposits can hide a third of a drop.

Methodology and sources

• Daily bear-market table: S&P 500 price index daily closes, 1927–October 2026 (S&P Dow Jones Indices data, downloaded from Yahoo Finance ticker ^GSPC; the 2016–2026 segment was cross-checked against FRED series SP500, which matches the 2020 and 2022 figures to one decimal). An episode starts at a record close, the trough is the lowest close before a new record, and recovery is the first close above the old record. Episodes are counted once even if they contain smaller rebounds. Our figures match independently published peak, trough and percentage values for 2000–02 (1,527.46 to 776.76, −49.1%), 2007–09 (−56.8%) and 1973–74 (−48.2%).
• Day-count statistics: 19,318 trading days from January 3, 1950 to October 14, 2026; a record close is any close at or above all earlier closes since 1927.
• Annual data: S&P 500 total returns (with dividends) and 10-year US Treasury bond returns, 1928–2025, from Aswath Damodaran's "Historical Returns on Stocks, Bonds and Bills" dataset (NYU Stern, updated January 2026). The 60/40 portfolio is rebalanced to 60/40 at the start of each year, ignoring costs and taxes. Calmar-style ratio = annualized return ÷ absolute max drawdown on year-end data.
• Derived figures: recovery gains = 1 ÷ (1 − loss) − 1; years to recover = ln(1 ÷ (1 − loss)) ÷ ln(1 + r). Median and other counts are computed from the table above.
• Limits: past drawdowns do not predict future ones. Index figures carry no fees or taxes. This is a measurement guide, not investment advice.