Stock market crashes in history: the biggest falls, rallies and flat decades of stocks, gold, bonds and crypto
The worst crashes of the S&P 500, gold, bonds, Bitcoin, Ethereum and Solana since 1928: depth, speed, time underwater, the biggest rallies and flat decades.
Every market people invest in has crashed. The S&P 500 lost 84% from 1929 to 1932, even with dividends. Gold lost almost two thirds after 1980 and did not see its old price again for 26 years. Long US government bonds, the "safe" part of many portfolios, fell 44% from 2020 to 2023 and have still not recovered. Bitcoin, Ethereum and Solana have each fallen more than 80% at least once.
This guide puts the biggest crashes of six markets side by side, measured the same way on daily closing prices: how deep each fall went, how fast it happened, how long investors waited to get back to even, and what caused and ended it. It then looks at the other side, the biggest rallies, and at the long stretches when markets went nowhere. The goal is a complete picture of what owning these markets has meant in history. It is education, not investment advice.
The data: six markets, up to 98 years
- S&P 500: the 500 largest US companies, with dividends reinvested (total return), daily from January 1928.
- Gold: US dollars per ounce from 1968 (monthly averages until 1986, daily since 1987). Before 1968 the price was fixed at $35.
- Long US Treasuries: a 20-year government bond with coupons reinvested, daily from 1962. It is rebuilt from the Federal Reserve's daily yield data, because bond funds did not exist that long; it tracks the TLT bond fund closely since 2002.
- Bitcoin from August 2011, Ethereum from August 2015 and Solana from April 2020.
All six end in mid-November 2026. Where US consumer prices are available, we also show the fall after inflation. Daily closes do not show moves within a day, so intraday crashes look smaller here than on a minute chart.
The five biggest crashes of each market
| Market | #1 | #2 | #3 | #4 | #5 |
|---|---|---|---|---|---|
| S&P 500 (with dividends) | −84% 1929–32 | −55% 2007–09 | −47% 2000–02 | −45% 1973–74 | −34% 2020 |
| −63% 1980–99 | −44% 2011–15 | −40% 1974–76 | −30% 2008 | −27% 2026 (ongoing) | |
| −44% 2020–23 (ongoing) | −26% 1980–81 | −24% 1979–80 | −22% 2008–09 | −19% 1987 | |
| −85% 2013–15 | −83% 2017–18 | −81% 2011 | −77% 2021–22 | −71% 2013 | |
| −94% 2018 | −79% 2021–22 | −77% 2015 | −68% 2025–26 (ongoing) | −67% 2016 | |
| −96% 2021–22 | −76% 2025–26 (ongoing) | −75% 2020 | −58% 2021 | −46% 2020 |
Peak to lowest daily close. Gold before 1987 uses monthly averages: from the daily high of about $850 in January 1980, the fall to 1999 was close to −70%.
Three things stand out. First, no market escaped: the mildest record, the 44% loss in long Treasuries, is still a loss that takes a doubling-plus-a-bit to recover. Second, crypto falls are in a different league, regularly 70% to 96%. Third, the size of a fall says little about how long it hurts. That is the next section.
How long investors waited: time underwater
A 50% fall needs a 100% gain to get back to even, and that gain can take years. Here is how long each market's deepest fall kept buyers from the top underwater.
| Deepest fall | Peak to low | Back at the old high | Time underwater | After inflation |
|---|---|---|---|---|
| S&P 500 1929–32 | 2.7 years | Jan 1945 | 15.3 years | 7.1 years |
| Gold 1980–99 | 19.6 years | May 2006 | 26.3 years | 44.8 years |
| Long Treasuries 2020–23 | 3.6 years | not yet | 6.7 years so far | not yet |
| Bitcoin 2013–15 | 13 months | Feb 2017 | 3.2 years | – |
| Ethereum 2018 | 11 months | Jan 2021 | 3.0 years | – |
| Solana 2021–22 | 14 months | Jan 2025 | 3.2 years | – |
The S&P 500 is the famous case. On price alone it needed 25 years to regain its 1929 high, until 1954. With dividends reinvested it needed 15 years. After inflation it needed only 7, because consumer prices fell by about a quarter in the early 1930s. Our guide to the 1929 crash covers that period in detail.
Gold is the more painful example and the less known one. An investor who bought in January 1980 did not see that price again until 2006, and in purchasing power not until 2025. That is 45 years. Gold has no dividend or interest to soften a long wait.
Crypto falls were deeper but shorter. Each of the deepest Bitcoin, Ethereum and Solana falls was made good within about three to four years. That is history, not a promise: the current falls of Ethereum (−68% from August 2025) and Solana (−76% from January 2025) had not recovered at the end of our data.
Fast crash or slow bleed? The shape of a fall
Two falls of the same depth can feel completely different. A crash that takes a month is over before most people react. A decline that grinds on for three years tests patience every week. We gave every fall one label, with fixed rules:
- Flash crash: peak to low in 14 days or less.
- Fast crash: peak to low in 15 to 92 days.
- Crash then grind: longer, with at least half of the fall in one 20-day stretch near the start.
- Grind then crash: longer, with at least half of the fall in one 20-day stretch near the end.
- Slow bleed: longer than 9 months, no single 20-day stretch holding half of the fall, and volatility not much above normal.
- Stair-step decline: several violent legs down, none of them dominant.
| Shape | Examples among the 30 biggest falls |
|---|---|
| Fast crash | S&P 500 2020 (−34% in 33 days), Bitcoin 2011 and April 2013, Ethereum 2015, Solana 2021 |
| Slow bleed | S&P 500 2000–02 and 1973–74, gold 2011–15, Treasuries 2020–23, Bitcoin 2013–15, 2017–18 and 2021–22 |
| Stair-step decline | S&P 500 1929–32 and 2007–09, Ethereum 2021–22, Solana 2020 |
| Crash then grind | Gold 1980–99 and 2026, Treasuries 2008–09, Ethereum 2016 |
| Grind then crash | Gold 2008, Treasuries 1979–80 |
Only the COVID crash of 2020 was a true fast crash among the deepest S&P 500 falls: 34% in 33 days, with almost all of it in the worst 20 trading days, and back at the high five months later. The others took one to three years to reach the low. The 2007–09 fall shows a typical stair-step: a slow first year, then the violent weeks after the Lehman collapse in September 2008, then a final leg into March 2009 (Federal Reserve History).
Did the deepest falls start with a flash crash?
Mostly not. We marked every fast crash inside each fall, using one rule: a drop of at least four times the market's typical daily move within one to five trading days (4.7% for the S&P 500, 16% for Bitcoin). In 18 of the 30 falls such a crash came within the first month after the peak. But it was only the opening move: the median first fast crash made up 17% of the eventual fall. October 1929, for example, brought the most famous crash days in market history, yet the S&P 500 then fell for almost three more years (Federal Reserve History). The lesson: a sharp first drop does not tell you whether you are looking at a fast crash that heals in months or the start of a long decline.
False rebounds
Almost every long fall contained a rally that looked like the end. Inside the 1929–32 fall the S&P 500 rose 49% from November 1929 to April 1930 before falling much further. In 2008 it rose 25% from late November to early January. Bitcoin rose 85% in the spring of 2014 and Ethereum 122% in April 2018, both inside falls that went on to new lows. Such bear market rallies are a normal part of how markets fall.
What caused the biggest crashes, and what ended them
- S&P 500, 1929–32 (−84%): a credit-fuelled boom ended in the October 1929 crash, followed by bank failures and falling prices in the Great Depression (Federal Reserve History). It ended at the 1932 low.
- S&P 500, 2007–09 (−55%): the US housing bust turned into a global banking crisis. Rescue packages, near-zero interest rates and Fed bond buying ended it in March 2009 (Federal Reserve History).
- S&P 500, 2000–02 (−47%): the dot-com bubble burst, followed by a recession and accounting scandals (Britannica).
- S&P 500, 1973–74 (−45%, −54% after inflation): the oil embargo and rising inflation during the Great Inflation (Federal Reserve History).
- Gold, 1980–99 (−63%): gold had soared with 1970s inflation; then the Fed under Paul Volcker raised interest rates sharply, which made a metal that pays no interest expensive to hold (Federal Reserve History). The low came in August 1999, weeks before European central banks agreed to limit their gold sales (World Gold Council).
- Gold, 2011–15 (−44%): after the crisis years the US economy recovered and the Fed prepared to stop its bond buying; 2013 was gold's worst year since 1981 (Quartz).
- Gold, 2026 (−27%, ongoing): after rising above $5,000, gold fell sharply on 30 January 2026 when Kevin Warsh was nominated as Fed chair (ETF.com), and slipped below $4,000 in October.
- Long Treasuries, 2020–23 (−44%): inflation returned and the Fed raised its rate from near zero to over 5%. 2022 was the worst year on record for Treasuries (Fortune).
- Long Treasuries, 1979–81: Volcker's fight against inflation pushed long yields to about 15% in September 1981 (Federal Reserve History). After inflation, long bonds lost 57% from 1965 to 1981, a slower and larger bond bear market than any nominal fall.
- Bitcoin, 2013–15 (−85%): China barred its banks from bitcoin business in December 2013 (NPR), and in February 2014 the largest exchange, Mt. Gox, collapsed (Cointelegraph).
- Bitcoin and Ethereum, 2018 (−83% and −94%): the boom in new coin offerings (ICOs) collapsed (MIT Technology Review).
- Crypto, 2021–22: rising interest rates, then the failures of the Terra stablecoin, crypto lenders and the FTX exchange (Decrypt). Solana fell 96%, hit hardest because FTX and its trading arm were among its biggest backers (Cointelegraph).
Did other markets protect investors?
A crash is easier to sit through if something else in the portfolio holds up. Over the exact peak-to-low window of each big S&P 500 fall:
| S&P 500 fall | ||||
|---|---|---|---|---|
| 2020 (Feb–Mar) | −34% | −3% | +13% | −32% |
| 2007–09 | −55% | +25% | +22% | – |
| 2000–02 | −47% | +12% | +40% | – |
| 1973–74 | −45% | +138% | −5% | – |
Bonds protected stocks in 2000–02, 2008 and 2020, but fell with them in 1973–74 and again in 2022, when both lost at the same time. Gold soared in 1973–74 and rose in 2007–09, yet slipped in the March 2020 dash for cash, when investors sold even safe assets to raise money (New York Fed). Bitcoin fell almost as much as stocks in March 2020 and fell with them in 2022. Our guide to the stock–bond correlation shows why the bond cushion comes and goes.
The other side: the biggest rallies
Crashes get the headlines, but the biggest moves up are just as extreme, and they tend to happen in the same years.
Explosive short rallies
The biggest one-day gains of the S&P 500 all came deep inside crashes. The record is 15 March 1933, +16.9%, the first trading day after the bank holiday (The Motley Fool). On 13 October 2008 the index rose 11.6% when governments announced bank rescue plans (NPR). The best 20-day run was +63% in July–August 1932, right after the low.
| Market | Best day | Best 20 days |
|---|---|---|
| +16.9% (15 Mar 1933) | +63% (to 8 Aug 1932) | |
| +9.5% (28 Sep 1999) | +28% (to 5 Oct 1999) | |
| +6.8% (20 Oct 1987) | +21% (to 18 Dec 2008) | |
| +56% (28 Oct 2011) | +295% (to 23 Nov 2013) | |
| +40% (19 Mar 2017) | +300% (to 11 Feb 2016) | |
| +46% (9 Sep 2020) | +275% (to 13 Aug 2020) |
Most of these jumps were given back within weeks: four of the S&P 500's five biggest one-day jumps were fully or almost fully lost again within three months. Gold's best day came when European central banks agreed to cap their gold sales in September 1999 (CNN Money); the best day for long Treasuries was the day after the 1987 stock market crash (Federal Reserve History).
Bull runs: the opposite of a crash
To measure long rises we count a bull run from a low until a fall of a fixed size begins: 20% for the S&P 500, gold and Treasuries, and a larger threshold for crypto, scaled to its higher volatility (55% for Bitcoin, 62% for Ethereum, 68% for Solana).
| Market | Biggest run | Gain | Length | Shape |
|---|---|---|---|---|
| Oct 1946 – Dec 1961 | +959% | 15.2 years | Stair-step | |
| Oct 1987 – Mar 2000 | +796% | 12.4 years | Stair-step | |
| Aug 1976 – Jan 1980 | +514% | 3.4 years | Melt-up | |
| Nov 2022 – Jan 2026 | +226% | 3.2 years | Stair-step | |
| Sep 1981 – Dec 2008 | +2,446% | 27.2 years | Stair-step | |
| Jan 2015 – Dec 2017 | +11,094% | 2.9 years | Stair-step | |
| Dec 2016 – Jan 2018 | +20,359% | 1.1 years | V-shaped rebound | |
| Dec 2020 – Nov 2021 | +21,473% | 0.9 years | V-shaped rebound |
The S&P 500's great bull markets were long climbs with many pauses: the best five days made up only about 8% to 15% of the gain of each run. Gold's biggest run was a melt-up: almost half of the 1976–80 gain came in its last six months, ahead of the January 1980 peak (LBMA). The bond bull market from 1981 ran for 27 years as yields fell from about 15% (investingLive). The latest gold run was carried by central banks, which bought more than 1,000 tonnes a year in 2022, 2023 and 2024 (Kitco, World Gold Council data).
Stairs up, elevator down? We measured it
The old saying is that stocks take the stairs up and the elevator down. We counted the average number of calendar days each market needed for a 10% move, over all its rising and all its falling legs.
For the S&P 500 the saying holds: 139 days per 10% on the way up, 70 on the way down, so falls ran about twice as fast. Long Treasuries are similar (330 vs 164). Gold is the opposite: it rose faster than it fell (113 vs 184 days), because its big falls were slow bleeds and its big rises were sprints. In crypto, rises were as fast as falls (Bitcoin 19 vs 18 days) or faster (Ethereum 12 vs 18, Solana 10 vs 18).
The best days come right next to the worst
All of the S&P 500's 10 best days came while it was more than 20% below its high, in the middle of crashes. For Bitcoin it was 7 of 10. Many of the best days came within two weeks of one of the worst. This is why the often-quoted cost of "missing the best days" is real but misleading:
Leaving out the 10 best days cut the S&P 500's long-run return from 10.1% to 8.8% a year, and Bitcoin's from 77% to 48%. Leaving out the 10 best and the 10 worst days left it about where it was, because they sit in the same weeks. Nobody could have picked only one side. Our guide does market timing work? goes deeper.
Going nowhere: the longest flat periods
Crashes and booms are the dramatic part of history. The part that tests investors most quietly is the long stretch with no progress.
Flat decades
| Bought at the price of… | …not above it for good until | Years |
|---|---|---|
| S&P 500 (price only), Oct 1968 | Aug 1982 | 13.9 |
| S&P 500 (price only), Apr 1999 | Nov 2012 | 13.6 |
| S&P 500 after inflation, with dividends, Oct 1961 | Aug 1982 | 20.8 |
| Gold, Jan 1980 | Aug 2007 | 27.6 |
| Gold after inflation, Jan 1980 | Jan 2025 | 45.0 |
| Long Treasuries after inflation, Jan 1962 | Dec 1985 | 23.9 |
| Long Treasuries after inflation, May 2003 | Oct 2026 | 23.4 |
| Bitcoin, Apr 2021 | Aug 2026 | 5.3 |
The 1966–82 and 2000–13 periods are the famous flat decades of US stocks. Dividends softened them, but after inflation the S&P 500 went nowhere for about 20 years from 1961. Long bonds bought in 1962 lost to inflation for almost 24 years, through the Great Inflation (Federal Reserve History). The same happened again to bonds bought in 2003: after inflation they were back below their purchase value in October 2026.
Trading ranges
A tighter view looks for periods when the price stayed inside a band without a lasting breakout: at least two years (one for crypto) with the high and low within 1.5 times the market's typical yearly move, ending near where it started.
| Market | Longest completed range | Band | Net change | After inflation |
|---|---|---|---|---|
| Jun 1946 – Oct 1949 (3.4 years) | 33% | 0% | −21% | |
| Mar 1990 – Jun 1997 (7.2 years) | 27% | −8% | −26% | |
| Mar 1962 – May 1970 (8.2 years) | 17% | +1% | −22% | |
| Aug 2014 – Jun 2016 (1.8 years) | 241% | +1% | – | |
| Aug 2018 – Jul 2020 (2.0 years) | 334% | −3% | – |
The hardest ranges to hold were not always the longest. We also ranked them by a pain score: years in the range times the average depth of the dips inside it. For the S&P 500 the 1946–49 range ranks first on both: nine swings of 10% or more, six tests of its floor, and post-war inflation that took a fifth of its value while the price stood still. For gold, the 1990–97 range: seven years of nothing, ending in a breakdown toward the 1999 low. In crypto, even "sideways" meant a band of three to five times from low to high.
How much of history was spent rising, falling or sideways?
We sorted every 12-month period: rising if the market moved up more than half its typical yearly move (about 10% for the S&P 500), falling if it moved down that much, sideways otherwise.
The S&P 500 was rising in 57% of all 12-month periods, sideways in 29% and falling in 14%. Gold spent more time sideways (35%) and falling (20%). Bitcoin rose in 60% of its 12-month periods, but its 17% of falling years were much deeper than anything in stocks.
What the history teaches a beginner
- Expect the depth. Every one of these six markets has lost at least 44% at some point, and most have lost much more. Before owning a market, it helps to know its worst fall and ask whether you could hold through it. Our max drawdown guide explains how to read this number.
- Time underwater is the hidden cost. The size of a fall decides how much is lost; the shape decides how long it hurts. Gold buyers of 1980 waited 26 years; stock buyers of 1929 waited 15 even with dividends.
- The first crash is rarely the whole story. Sharp first drops were usually only the opening move of a longer decline, and big rebounds inside a fall often failed.
- Hedges change over time. Bonds and gold protected stocks in some crashes and fell with them in others. Our guide on how to prepare for a stock market crash lists what Treasuries, cash and gold did in each of the 12 big S&P 500 sell-offs since 1955.
- Inflation turns flat into a loss. A market that goes sideways while prices rise is losing purchasing power, as stocks and bonds did in the 1960s and 1970s.
- The best and worst days come together. That makes it very hard to step aside in a crash and still catch the recovery.
- Diversification across markets with different shapes is one way investors have dealt with this. Our free Portfolio Crash Test shows how a mix of assets would have held up in every big crash since 1972.
For single crashes in close-up, see our guides on the crash of 1929 and the 1987 crash; for crypto's fastest falls, the biggest crypto crashes, ranked. How the markets in this guide compare head to head is covered in gold vs S&P 500, stocks vs bonds and Bitcoin vs S&P 500. For where the S&P 500, Nasdaq-100, gold and Bitcoin stand against their highs at the last close, with the rules of this guide applied, see Is the stock market crashing?.
See the falls live
The Research Terminal shows each market's current fall from its high, its trend and its history of drawdowns, next to how a systematic strategy compares with buy & hold, for six core markets in the free account. You can check the S&P 500, gold, long Treasuries and Bitcoin market pages for each market's own crash history.
Method: daily closing prices; drawdowns from the previous all-time high; rules for shapes, rallies and ranges as described above. Long Treasuries before 2002 are a reconstruction from Federal Reserve yield data; gold before 1987 uses monthly averages. Past falls and recoveries say nothing certain about the future. Educational research, not investment advice.
Frequently asked questions
What was the biggest stock market crash in history?
For US stocks it was 1929–32: the S&P 500 fell about 86% in price and 84% with dividends reinvested, from September 1929 to June 1932. It took until 1954 to regain the 1929 high in price, and until 1945 with dividends. The fastest big crash was in 2020, when the S&P 500 fell 34% in 33 days.
How long does it take the stock market to recover from a crash?
It depends on the depth and the cause. With dividends reinvested, the S&P 500 needed about 5 months after the 2020 crash, 3.5 years after 1973–74, 4.9 years after 2007–09, 6.6 years after 2000–02 and 15 years after 1929. Inflation can make the wait much longer: after 1973–74 the real recovery took 12 years.
How long do bear markets last?
Among the five deepest S&P 500 falls since 1928, the time from peak to low ranged from 33 days (2020) to almost 3 years (1929–32 and 2000–02). The full time underwater, from peak back to the old high, ranged from 6 months to 15 years. Gold and long bonds have had even longer bear markets.
What is the worst bond bear market in history?
In our data from 1962, long US Treasuries (20-year, interest reinvested) fell 44% from March 2020 to October 2023, as inflation returned and the Fed raised rates; they had not recovered by November 2026. After inflation, the 1965–1981 bond bear market was about as bad: −57% in purchasing power.
How much did gold fall after 1980?
Gold fell from its January 1980 peak (a daily high near $850, monthly average $675) to about $252 in August 1999, a fall of 63% from the monthly average and about 70% from the daily high. It regained the January 1980 monthly price in 2006 and, after inflation, only in 2025.
What were the biggest Bitcoin crashes?
Bitcoin fell 85% from December 2013 to January 2015, 83% from December 2017 to December 2018, 81% in 2011, 77% from November 2021 to November 2022 and 71% in April–July 2013 (daily closes). Each of these was recovered within about one to three years.
Do stocks fall faster than they rise?
Yes, on average. Over all rising and falling legs since 1928, the S&P 500 needed about 139 days per 10% rise and 70 days per 10% fall, so falls ran about twice as fast. Gold showed the opposite pattern, and crypto rose about as fast as it fell or faster.
Did bonds or gold protect stocks in crashes?
Sometimes. Long Treasuries rose in the 2000–02, 2007–09 and 2020 stock crashes but fell in 1973–74 and in 2022. Gold rose strongly in 1973–74 and in 2007–09 but slipped in the March 2020 crash. No hedge worked in every crash.
Cite or link this guide
Writing about this topic? You are welcome to quote the figures with a link to this page. Click a box to copy.
Citation (APA)
Hamann, H. (2026). Stock market crashes in history: the biggest falls, rallies and flat decades of stocks, gold, bonds and crypto. InvestWealth.com. https://www.investwealth.com/learn/guides/market-crashes-history
Link (HTML)
<a href="https://www.investwealth.com/learn/guides/market-crashes-history">Stock market crashes in history: the biggest falls, rallies and flat decades of stocks, gold, bonds and crypto</a> (InvestWealth.com)
Know the falls before the next one comes
- Drawdown monitor: how far six core markets are below their highs, daily
- The four deepest falls of each market and the days to recover
- Strategies: a rule-based approach against buy & hold through the same falls
Related guides
How much Bitcoin a portfolio can carry, worked out from the falls rather than the returns: Bitcoin's drawdowns since 2017, how it behaved when stocks…
Gold vs bonds: which protects a portfolio better in a crash?Gold rose 138% in 1973–74 when bonds failed, then fell 46% in 1980–82. Every big sell-off since 2007 and 150 years of data: when each hedge worked.
How to prepare for a stock market crash: what protected a portfolio in 12 crashes since 1955How to prepare for a stock market crash and hedge a portfolio or 401(k): what Treasuries, cash and gold did in the 12 big S&P 500 sell-offs since 195…
Daily, monthly or yearly compounding, with monthly contributions.
Investment calculator with dividend reinvestmentReinvested vs paid out, plus the real S&P 500 history.
Portfolio Crash TestHow your mix would have held up in every big crash since 1972.
Educational research, not investment advice. Disclaimer.