What Is a Stock Market Crash? Causes, Circuit Breakers and What a Crash Looks Like on a Chart

What is a stock market crash? It's a sudden, steep drop in stock prices across most of the market, usually measured by the S&P 500, the Dow or the Nasdaq, over a day to a few days rather than months, and driven largely by panic selling. No regulator sets an official number.

I trade momentum, so my interest in crashes is narrow. I care what they do to the chart in front of me and to the stop under my position. The first half covers the definition, causes, circuit breakers and history, with every figure traced to a primary source. The second half is the chart side, which most crash explainers skip because they're written for long-term investors.

Quick Answer

The halt rules come from Investor.gov's glossary entry on stock market circuit breakers, and the bear market definition from its bear market glossary entry. Index levels come from Federal Reserve History, Britannica and FRED's daily closes, linked where they're used.

Stock Market Crash Definition: Is There an Official Number?

No. There's no official stock market crash definition. Investor.gov, the SEC, FINRA and the NYSE don't set a crash threshold anywhere I could find, and Investor.gov's glossary has no entry for the word at all. It describes a sudden, steep fall across most of the market, and writers attach their own numbers to it.

Those numbers disagree. Explainers I read gave 10% in a single day, 10% or more over several days, 10% to 20% in a single week, and 20% or more over a few days, each presented as the definition. They're conventions. If you want one working line: a crash is commonly a double-digit drop in a broad index over a few sessions, and the speed matters more than the exact percentage.

Breadth matters too. One stock dropping 40% on bad earnings is a terrible day for that stock. A crash is when the index falls because most of its members fall together, which is why crash coverage quotes the Dow, the Nasdaq Composite or the S&P 500, the index most US market data is measured against, instead of a single name.

Stock market crash vs correction vs bear market

The three words measure different things. Correction and bear market are about depth. Crash is about speed. A crash can be the opening move of a bear market, or a sharp drop the market climbs back from. The 20% rule and how corrections differ from bear markets already live in the bull vs bear market explainer, so the table mostly adds the crash column.

For a trader the row that matters is speed. A correction or a bear market gives you weeks of lower highs to notice what's happening. A crash squeezes that into a few sessions, and that compression is what breaks stops and position sizes.

What Causes a Stock Market Crash?

A crash usually needs a market that has run up on optimism or borrowed money, plus something that flips buyers into sellers all at once. The trigger is different each time, and the selling that follows works the same way.

How does a stock market crash work?

Prices move when one side runs out. On a normal day, a seller who wants out finds a buyer a few cents away, at the bid side of the spread. In a crash, sellers line up faster than buyers show up, so each trade prints lower than the last. Federal Reserve History's account of the 1987 crash describes exactly that: on Black Monday, "the number of sell orders vastly outnumbered willing buyers near previous prices, creating a cascade." Falling prices then set off margin calls and automated selling, which adds sell orders, which pushes prices lower again. That loop is the difference between a crash and an ordinary bad day.

The usual causes

• A long run-up with speculation. Crashes tend to follow stretches of rising prices and confident buying. By late August 1987, per Federal Reserve History, the Dow had gained 44% in seven months. The dot-com crash followed a run in technology stocks, and 1929 followed the speculative boom of the late 1920s.
• Borrowed money. Buying on margin magnifies a fall. In the 1920s, per Federal Reserve History's essay on the 1929 crash, buyers "put down a fraction of the price, typically 10 percent, and borrowed the rest." When prices drop, the lender wants more collateral, and anyone who can't post it gets sold out. The margin call walkthrough shows how that trigger is calculated today.
• A shock. A pandemic, a war, a disaster or a sudden policy change can reprice everything at once. The 2020 crash was the market repricing a pandemic in a few weeks.
• Tighter money.Higher rates make borrowing to buy stocks more expensive. In August 1929 the New York Fed's discount rate reached 6%, two months before the crash.
• Automated selling.In 1987, a hedging approach called portfolio insurance sold more as prices fell. Federal Reserve History says it "accelerated the crash's pace as initial losses led to further rounds of selling."

Can you see a crash coming?

Not reliably. Run-ups, heavy margin borrowing and stretched valuations can sit in place for years without a crash, and the trigger is often something nobody had on their list. The longer version, including what AI models can and can't forecast, is in my piece on whether AI can predict the stock market. Short version: a chart shows what has already happened.

What happens to the economy after a crash?

Sometimes a recession follows and sometimes nothing much does. Lost wealth and shaken confidence can slow spending and lending, and 1929 ran into the Great Depression. But Federal Reserve History notes that the sharp losses of Black Monday "were not followed by an economic recession or a banking crisis." The central bank usually steps in either way. After 1929 the New York Fed bought government securities and lowered the discount rate. And on Sunday, March 15, 2020, the Federal Reserve cut interest rates to nearly 0%, per the SEC-filed circuit breaker review covered in the next section.

How Do Stock Market Circuit Breakers Work?

Stock market circuit breakers are automatic, market-wide trading halts triggered by a single-day drop in the S&P 500. There are three levels: 7%, 13% and 20% below the prior day's close. Levels 1 and 2 halt all trading for 15 minutes if they hit before 3:25 p.m. ET. Level 3 ends trading for the rest of the day, whenever it hits.

A few details most explainers leave out. The trigger levels are calculated each day from the previous close, so they're fixed points for the session and don't move with the morning high. Levels 1 and 2 can each halt trading only once a day, per the circuit breaker working group report filed with the SEC, so a market that drops through 7%, reopens and keeps falling won't halt again until 13%. Investor.gov calls them "cross-market trading halts", which means every venue stops together. There's nowhere else to sell the stock during one.

Where circuit breakers came from

They were a response to Black Monday. The same report says the cash equities rules for market-wide circuit breakers "were first adopted in 1988", and that in 2012 all US stock exchanges and FINRA amended them on a pilot basis. The current S&P 500-based 7%, 13% and 20% levels come from that 2012 amendment. Some explainers say the SEC created the 7/13/20 levels right after 1987. The levels are newer than that, and the rules belong to the exchanges and FINRA, with SEC approval.

The four halts of March 2020

Level 1 halted the whole US market four times in March 2020. The working group's review of those days gives the exact second the S&P 500 crossed 7% down.

Two things stand out. On March 16, selling was so heavy that the halt hit at 9:30:01 a.m., and the report says most S&P 500 stocks hadn't even finished their opening auction. And on March 18 the index crossed 7% down after lunch but closed down 5.18%. A circuit breaker fires on the intraday drop, and the daily close can look much milder than the day felt.

Individual stocks have their own brakes. Under Limit Up-Limit Down, each stock has a price band and trades can't print outside it. Per Investor.gov, a stock that hits its band and doesn't move back inside within 15 seconds pauses for 5 minutes. The band sizes are in the section on when a penny stock gets halted, and the guide to volatility in stocks covers how those pauses fit with spreads and slippage on a fast tape.

The Biggest Stock Market Crashes in History

Five episodes show up in nearly every crash explainer: 1929, 1987, the dot-com bust, 2008 and 2020. Popular explainers misquote several of them (the 1987 drop shows up as 22%, 23% and even 25%), so the figures below stick to Federal Reserve History, Britannica and FRED. The last column is the gain the index needed just to get back to where it started.

Read the last column before the others. The deeper the drop, the more absurd the climb back gets, and that's before you count the years it took.

1929: Black Thursday, Black Monday, Black Tuesday

The 1929 crash ran over several sessions. On Black Thursday, October 24, a record 12.9 million shares traded, but the Dow closed down only six points after major banks and investment companies bought up large blocks of stock, per Britannica's account of the 1929 crash. Then the selling took over. On Black Monday, October 28, the Dow fell nearly 13%, and on Black Tuesday, October 29, it fell nearly 12% more. Over those four business days, by Britannica's count, it went from 305.85 to 230.07, about a quarter of its value.

And that was only the start. The Dow kept sliding until July 8, 1932, when it closed at 41.22, 89% below its September 3, 1929 peak. It didn't get back to the 1929 high until November 23, 1954, per Federal Reserve History.

1987: Black Monday

On October 19, 1987, Federal Reserve History says, the Dow "crashed at the opening bell and eventually finished down 508 points, or 22.6percent." The Friday before, it had already lost 4.6%. Portfolio insurance sped up the fall. The bounce was fast too: in two sessions the Dow won back 57% of the Black Monday loss, and US stocks passed their pre-crash highs less than two years later.

2000 to 2002: the dot-com crash

The dot-com crash happened in slow motion. The Nasdaq Composite closed at 5,048.62 on March 10, 2000, and kept falling for two and a half years, to 1,114.11 on October 9, 2002, a 77.9% decline in FRED's Nasdaq Composite series. It didn't close above the 2000 peak again until April 23, 2015. No single day defines it. It makes every list because of the depth.

2008: the financial crisis

The 2008 crash came out of housing and credit. Per Federal Reserve History's essay on the Great Recession, the S&P 500 fell 57% from its October 2007 peak to its March 2009 trough, and the recession itself ran from December 2007 to June 2009. The recovery dates other sites give for 2008 disagree by years, so I've left them out.

2020: the COVID crash

The 2020 crash was the quickest 30%-plus fall on this list. The S&P 500 dropped 33.9% from 3,386.15 on February 19 to 2,237.40 on March 23, in 23 trading sessions, per FRED's S&P 500 series. The Dow fell 37.1% from its February 12 peak, per FRED's Dow Jones series. The recovery was quick as well, and the date depends on the index: the Nasdaq Composite closed above its February peak on June 8, the S&P 500 on August 18 and the Dow on November 16, 2020.

Recovery: what history shows and what it doesn't

Every crash in the table recovered eventually. The wait ranged from about six months for the S&P 500 in 2020 to 25 years for the Dow after 1929, and nothing in the data says the next one picks from that range. Losses on a position you still hold aren't realized until you close it, which is true and gets stretched a lot. Some crash explainers turn it into "you don't lose money unless you sell." A trader with a stop, a margin loan or a position in a stock that never comes back loses real money, fast.

What helps most is staying calm and knowing your exit before the open, so that on a crash day you're executing a plan instead of inventing one at the worst prices of the year.

What Does a Stock Market Crash Look Like on a Chart?

On a chart, a crash shows up as gap-down opens, candles several times their normal size, volume far above average, support levels breaking one after another, and violent relief rallies that don't mark the bottom. They usually arrive together, and they describe a crash while it's happening rather than predict one.

Gap-down opens

When bad news lands overnight, the first trade of the day prints far below the prior close and nothing trades in between. On February 24, 2020, the S&P 500 opened 2.4% below the previous close and ended the day down 3.4%, per the working group report. On March 16 the opening drop was steep enough that the Level 1 halt hit one second after the bell. The breakdown of breakaway, runaway and exhaustion gaps is useful here, because a gap that opens a move and a gap that ends one look alike on the first candle.

Wide-range candles

Normal sessions have small candles. Crash sessions don't. On March 3, 2020, the market's trading range was 5.2%, and the report notes the average daily move over the first three weeks of February had been 0.7%. That's a range more than seven times the recent norm. Long red bodies that close near their lows stack up (a bearish marubozu candle is the extreme version), and range-based indicators like average true range jump with them.

Volume spikes

Volume explodes. On February 28, 2020, US stocks traded 19.375 billion shares, at the time the second most active day in history, according to the same report. In 1929, Black Thursday's 12.9 million shares was a record and Black Tuesday topped 16 million. On an intraday chart this shows up as relative volume far above normal on almost every name at once, which is one way to tell a market-wide event from a single stock's bad day.

Support breaking, one level after another

Levels that held for months give way in a session or two. The buyers who defended a level last time don't show up, and the broken level often caps the next bounce. If you mark support and resistance zones on a crash chart, expect to redraw them a lot.

Violent relief rallies

Some of the biggest up days land in the middle of a crash. The S&P 500 rose 9.29% on March 13, 2020, 6.00% on March 17 and 9.38% on March 24, and the first two came before the March 23 low. In 1987 the Dow won back 57% of the Black Monday loss in two sessions. A huge green candle after a string of red ones looks like a bottom and may not be one. The guide to bull traps and bear traps covers how those reversals fail.

Momentum traders read the same features on ordinary days (gaps, range expansion, relative volume), which is why the momentum trading strategy playbook leans on them. In a crash they show up on nearly every chart at once, which tells you what kind of day it is but not where the low is.

What Happens When the Stock Market Crashes? Your Stops, Size and Margin

For a trader, a crash is mostly a risk problem. Gaps jump stops, wider ranges throw off position sizes, margin turns losses into forced exits, and the math of getting back to even gets brutal. Not losing more is the first job.

If you're a long-term investor, the standard advice (stay diversified, rebalance, don't sell in a panic, ask a professional if you're unsure) is easy to find elsewhere. This section is for people holding trades with stops.

The recovery math

Losses and gains aren't symmetric. A 33.9% drop, the S&P 500's 2020 fall, needs a 51.3% gain to get back to even. A 57% drop like 2008 needs about 133%. The Dow's 89% fall from 1929 to 1932 needed about 809%. The formula is gain needed = 1 / (1 - loss) - 1, and the history table applies it to every row. That asymmetry is the whole case for cutting a loss early in a crash instead of hoping.

Don't catch a falling knife

Buying the first huge red candle means buying before any structure exists. There's no higher low and no level that has held, so there's nothing to put a stop under. The relief rallies above show the risk: a big green day can be followed by a lower low the next session. The difference between buying a dip and catching a falling knife comes down to waiting for structure, and in a crash that wait can be long.

Why your stop can fill far past its price

A stop-loss is an order. A standard stop becomes a market order once its price is reached, so the fill is whatever the next trade is. Through a gap-down open that can be far below your stop. During a market-wide halt nothing trades at all, and if your stop triggers when trading resumes, it fills wherever the market reopens. A stop-limit avoids the bad fill by risking no fill at all. The explainer on what happens when price gaps past your stop walks through what that costs with numbers.

Wider ranges mean smaller size

If you risk a fixed dollar amount per trade, the stop distance sets your share count. In a crash, ranges widen several times over, so a sensible stop sits further away and the same dollar risk buys fewer shares. Keeping your usual size with a wider stop means risking several times your usual amount on one trade. The walkthrough on sizing a position from risk per trade covers the arithmetic, and the day trading risk management rules include the daily loss limit, which matters most on days like these.

Margin turns a crash into a forced exit

The 1929 buyers who put down 10% learned this the hard way. When prices fall, the lender wants more collateral, and if you can't post it your position gets sold, often near the worst prices of the move. It works the same way today, and a margin call doesn't wait for a recovery. The guide to when a margin call triggers shows the math, and the margin account explainer covers why your losses can run past what you deposited.

What a Chart Grade Can and Can't Tell You in a Crash

SnapPChart grades one chart screenshot. You upload the image, it reads the candles, EMAs, VWAP, MACD and volume in it, and grades the momentum continuation setup, long or short. On full analyses a setup that qualifies comes with an entry, a stop with its reasoning, and targets. That image is the whole input. It doesn't read the news, the VIX, halt status or live data, and it doesn't predict crashes or tell you one has ended.

On a crash chart it reads the established trend from the screenshot, not just the last few candles. A clean downtrend with a lower high and a pullback can be graded as a short-side continuation setup. A long only gets graded if the screenshot actually shows one forming. When the move is parabolic or the structure can't be read, it skips the trade and tells you so. A skip on a crash day is a useful answer: it means the picture doesn't support a trade with a defined stop. Grading before you trade is mostly about the trades you don't take.

For a wider look at what AI can and can't read off a chart, there's an overview of AI chart analysis and a broader piece on where AI fits in day trading.