History of stock market crashes: every major US drawdown compared

Stock markets fall hard every so often. Since 1929 the US market has been through a long list of crashes and bear markets: some over in weeks, some lasting years, one that took a quarter of a century to repair on price alone. Each had its own trigger, and each felt, at the time, as if it might never end. Seen together, they are less mysterious. The same handful of forces shows up again and again, and so do the same mistakes.

This guide is the hub for the market-history pages in the Finabase Learn library. It compares the major US drawdowns in one table, built only from well-documented published closing values, shows the post-1993 record on real SPY data, sorts the episodes by cause, and draws out what the record does and does not tell an investor. Each episode has its own page with the full story; the links are in each section below.

Crash, correction, bear market: the words

There is no official definition of a crash. People use the word for a sudden, steep fall over a day or a few days, such as October 1929, October 1987 or March 2020. The more formal conventions apply to the size of a decline from a recent closing high:

• A pullback is a modest dip, often 5% to 10%.
• A correction is a decline of 10% to 20%.
• A bear market is a decline of 20% or more.

A drawdown is the general term for any peak-to-trough fall, and it is the measure used in the tables here. Those conventions, and how bull markets are dated, are covered in Bull and bear markets: definitions, corrections and history. On this page, "crash" means one of the famous episodes, and the numbers always measure the whole decline from the closing peak to the closing low.

The major US drawdowns compared

The table lists the deepest US bear markets of the past century, plus 1987 and 2020 for their speed. Index values are daily closes from S&P Dow Jones Indices data as published in Yardeni Research's bear-market tables, Dow Jones history for 1929, and Nasdaq for the Composite. All figures are price only: no dividends, no inflation adjustment.

Several smaller bear markets are left out to keep the table to well-known episodes, for example 1946, 1968–70 and 1980–82, with S&P 500 declines of 26.6%, 36.1% and 27.1%. The full list is in the Yardeni tables cited below.

Two cautions about reading the last column. First, price-only recovery overstates the wait, sometimes a lot, because dividends keep paying through a bear market and can be reinvested at low prices. Second, it understates the wait for anyone who measures in purchasing power: the 1973–74 decline came with high inflation, so the real loss lasted much longer than the price chart shows.

Thirty years on real data

Our own price history does not reach back to 1929, but the S&P 500 ETF, SPY, has traded since 1993, and its monthly closes cover four of the bear markets in the table.

The chart uses a log scale, on which equal percentage moves look equal in height, so a 50% fall in 2008 is drawn the same size as a 50% fall in 1990s prices. Two things are visible at a glance. The 2000–02 and 2007–09 bear markets were deep and slow, and SPY spent most of 2000 to 2013 below its 2000 peak. The 2020 and 2022 declines, by contrast, look like notches. Monthly closes also understate fast crashes: on daily closes the 2020 fall was 33.9%, but SPY's monthly closes show only about 20%, because March's worst days were between month-ends.

Single-day crashes

A different list ranks the worst single days. These are rare, and most cluster in a few episodes. The selection below shows the Dow's largest one-day percentage falls from the episodes covered in this library; it is not a complete ranking.

Big down days tend to arrive in clusters inside bear markets, often next to big up days. Some of the Dow's largest one-day percentage gains came in the same weeks as its worst days: October 30, 1929 (+12.3%), October 21, 1987 (+10.1%), October 13, 2008 (+11.1%) and March 24, 2020 (+11.4%). That is one reason trying to step aside for the worst days usually means missing the best ones too.

The episodes in brief

1929–1932: the Great Crash

A credit-fueled boom ended in two days of panic in late October 1929, then turned into an 89% decline over nearly three years as thousands of banks failed and the money supply shrank. The reforms that followed created the SEC, disclosure rules, federal margin limits and deposit insurance. Full story: The crash of 1929: causes, the 1929–32 decline and lessons.

1973–1974: inflation and an oil shock

The S&P 500 fell 48.2% from January 1973 to October 1974 as inflation surged, the 1973 oil embargo quadrupled oil prices, and the economy entered recession. The era's favorite large growth stocks, priced at very high multiples, fell especially hard. Because inflation stayed high for the rest of the decade, the real, inflation-adjusted loss lasted far longer than the price recovery in 1980 suggests. (This library does not yet have a separate page on 1973–74.)

1987: Black Monday

A 33.5% S&P 500 decline from August to December 1987, including the Dow's 22.6% fall on October 19, driven largely by market mechanics such as portfolio insurance. The economy kept growing and the market regained its peak by mid-1989. It led directly to market-wide circuit breakers. Full story: Black Monday 1987: the 22.6% crash and why circuit breakers exist.

2000–2002: the dot-com bust

The internet boom took the Nasdaq Composite to 5,048.62 in March 2000; it then fell 78% by October 2002, while the broader S&P 500 fell 49%. Overvaluation, not credit, was the core problem, so rate cuts could not stop the slide. Full story: The dot-com bubble: the 1995–2002 boom and the 78% Nasdaq fall.

2007–2009: the financial crisis

A housing bust exposed leveraged holders of mortgage securities. Lehman Brothers' failure in September 2008 turned it into a global panic, and the S&P 500 fell 56.8%, the deepest decline since the 1930s. TARP, near-zero rates and bond buying stabilized the system. Full story: The 2008 financial crisis: housing, Lehman, TARP and the 57% fall.

2010: the Flash Crash

Not a bear market, but a warning about market structure: on May 6, 2010 the Dow fell about 9% in minutes and mostly recovered by the close, after an automated futures sell program hit a thin market. It produced the Limit Up-Limit Down bands and today's circuit-breaker design. Full story: The 2010 Flash Crash: what happened on May 6 and what changed.

2020: the COVID crash

The fastest bear market on record: 16 trading days from the February 19 record to a 20% decline, 33.9% in 33 days, four circuit-breaker halts, and a record VIX close. An exceptionally large and fast policy response helped the S&P 500 regain its record by August. Full story: The COVID crash of 2020: the fastest bear market and recovery.

2021: meme-stock mania

A bubble and bust in a handful of stocks rather than the whole market. GameStop rose about 2,700% intraday in January 2021 and lost most of it within weeks, while brokers restricted buying because of clearing collateral demands. Full story: Meme stock mania of 2021: GameStop, AMC and the SEC report.

2022: the rate-shock bear market

Inflation at a 40-year high and the Fed's fastest hikes in decades pushed the S&P 500 down 25.4% and long-term Treasuries down even more, so stocks and bonds fell together. Stocks regained their record in January 2024. Full story: The 2022 bear market: inflation, Fed hikes and falling bonds.

What the crashes have in common

Leverage turns declines into cascades

In 1929 it was stock bought on margin. In 1987 it was portfolio-insurance selling, a form of mechanical, forced selling. In 2008 it was banks and investment firms holding mortgage securities with borrowed money. In each case, falling prices forced sales, and forced sales pushed prices lower. The declines that went deepest were the ones where the financial system itself was leveraged. See Buying on margin: how margin loans, interest and leverage work and Margin calls: when they trigger, how to calculate them, what to do for the individual-investor version.

Valuation sets the depth

The two slowest recoveries in the table, after 1929 and after 2000 for the Nasdaq, both started from extreme valuations. When prices assume decades of strong growth, a disappointment can take a long time to work off, even if the economy recovers quickly.

Policy shapes the length

The contrast between 1929–32 and 2008–09 or 2020 is mostly about the response. In the early 1930s the Federal Reserve let banks fail and money contract. In 2008 and 2020 central banks and governments moved quickly to keep credit flowing, and the market bottomed long before the economic data did.

Liquidity disappears when it is needed most

1987, 2010 and March 2020 all featured moments when buyers simply stepped away, and prices gapped. Circuit breakers, LULD bands and central-bank backstops all exist because of those moments. Market-wide circuit breakers: the 7%, 13% and 20% halt rules explains the current rules.

What differs

No two crashes had the same cause. The trigger was a credit bust in 1929 and 2008, a technology bubble in 2000, inflation in 1973 and 2022, market mechanics in 1987 and 2010, and a pandemic in 2020. The speed also varies enormously, from 33 days to 34 months to the trough. A rule drawn from one episode, such as "crashes recover in months" from 2020 or "they take a decade" from 2000, fails on the next. The table's honest message is a range.

The arithmetic of recovery

Losses and gains are not symmetrical. After a decline, the gain needed to get back to even is larger than the loss.

This is why the deepest declines took so long to repair, and why limiting losses, through Diversification: why spreading your investments reduces risk across assets and through not borrowing to invest, matters more than chasing the last few points of return.

What the record means for investors

• Bear markets are normal. The table is not a list of freak events; counting the smaller ones it leaves out, the S&P 500 has averaged more than one 20% decline per decade since 1946. Planning for them is part of owning stocks.
• Time horizon is the main defense. An investor who needed the money within a year of the 2000 or 2007 peaks faced a very different outcome from one with 20 years to go. See Time horizon and risk tolerance: matching investments to goals.
• Index recoveries are not stock recoveries. Indexes drop failing companies and add new ones. Many individual stocks from 1929, 2000 and 2008 never regained their peaks.
• Selling in the panic has been costly. In 1987, 2009 and 2020 the lows came during the worst news. That pattern is history, not a guarantee.
• Steady investing changes the experience. Someone buying a fixed amount each month, as in Dollar-cost averaging vs lump sum: how DCA works and what data show, bought at low prices through every one of these declines.

How Finabase helps

The SPY monthly chart shows the S&P 500 ETF from 1993 to today on the same data as the long-run figure above, and the comparison view lays SPY next to the Nasdaq-100 ETF and long Treasuries to see how different assets behaved in each decline. The indices page shows where the S&P 500, the Dow and the Nasdaq stand today, and the VIX page shows the market's expected volatility. Our stored index history begins in November 2021, so for older episodes we use the ETFs, which carry their full history, and published index values for anything before 1993.

Frequently asked questions

What was the worst stock market crash in US history?

By depth, the 1929–1932 decline: the Dow fell 89.2% from September 3, 1929 to July 8, 1932. By single-day fall, October 19, 1987, when the Dow lost 22.6%.

How long does it take the stock market to recover from a crash?

It has varied widely. On price alone, the S&P 500 took about 6 months to regain its record after 2020, about 2 years after 1987 and 2022, about 5½ years after 2007 and about 7 years after 2000. The Dow took 25 years after 1929. Dividends shorten those waits.

How often do bear markets happen?

Using the 20% convention on daily closes, the S&P 500 has had more than a dozen bear markets since 1929, but they do not arrive on a schedule: there were two within three years in 2020 and 2022, and none between 2009 and 2020, when declines in 2011 and 2018 stopped just short of 20%.

What causes stock market crashes?

Usually a shock meeting a vulnerable market: high valuations, heavy borrowing, or fragile market structure. The shock has ranged from a credit bust to a pandemic. Crashes deepen when forced selling and disappearing liquidity feed on each other.

Can you predict the next crash?

Not reliably. Valuations, leverage and credit conditions say something about vulnerability, but the timing of the trigger has never been predictable. Every episode on this page surprised most investors when it began.

Summary

From the 89% collapse of 1929–32 to the 25% rate-shock decline of 2022, the major US bear markets have ranged from 33 days to nearly three years from peak to trough, and from six months to 25 years from peak to a new price high. The deepest combined high valuations with financial-system leverage; the shortest were sudden shocks met by a fast policy response. The single-day record still belongs to October 19, 1987. The history shows US indexes have recovered every time so far, but on very different timetables, which is why time horizon, leverage and diversification matter more than any attempt to call the next one.