How Often Does the Stock Market Drop 1%, 2% or 3% in a Day? Every S&P 500 Session Since 1950, by Decade, and What It Means for Your Stop-Loss

The S&P 500 closes down 1% or more on about one session in ten. Across the 19,331 trading days from January 3, 1950 to November 3, 2026, it happened 1,942 times, 10.0% of all sessions, which works out to 25.3 days a year. A 2% drop happened 444 times, 5.8 a year. A 3% drop 132 times, 1.7 a year. A 5% drop 30 times in 77 years. Up days of the same size are about as common: 2,090 sessions up 1% or more, 427 up 2% or more. One session in five moves at least 1% in one direction or the other. Those are the base rates. The rest of this page is how they changed by decade, what 2026 looks like against them, what tends to happen the day after a big drop, and what each number means if you run a stop-loss.

The count, by decade

Per-year rates, so decades with different session counts compare cleanly. The 2020s row is through November 3, 2026.

• 1950s: 15.6 days a year down 1% or more, 2.3 down 2% or more, 0.6 down 3% or more (2,510 sessions).
• 1960s: 13.3 days a year down 1% or more, 1.4 down 2% or more, 0.3 down 3% or more (2,489 sessions).
• 1970s: 26.1 days a year down 1% or more, 3.0 down 2% or more, 0.3 down 3% or more (2,526 sessions).
• 1980s: 26.5 days a year down 1% or more, 5.1 down 2% or more, 1.3 down 3% or more (2,528 sessions).
• 1990s: 22.3 days a year down 1% or more, 4.2 down 2% or more, 0.9 down 3% or more (2,528 sessions).
• 2000s: 43.2 days a year down 1% or more, 14.6 down 2% or more, 5.2 down 3% or more (2,515 sessions).
• 2010s: 24.6 days a year down 1% or more, 7.1 down 2% or more, 1.9 down 3% or more (2,516 sessions).
• 2020s: 33.3 days a year down 1% or more, 9.8 down 2% or more, 4.0 down 3% or more (1,719 sessions).

Two decades stand out. The 2000s, with the dot-com unwind and 2008, averaged 43.2 one-percent down days a year, nearly three times the 1960s. The 2020s so far, with the pandemic crash and 2022, run at 33.3, the second-highest on the list. The calmest decade was the 1960s at 13.3; the longest stretch in the whole record without a single 1% down day was 154 sessions, from 1963-03-01 to 1963-10-09.

The last fifteen years, year by year

Days down 1% or more, 2% or more, 3% or more, with the session count.

• 2012: 21 / 3 / 0 in 250 sessions.
• 2013: 17 / 2 / 0 in 252 sessions.
• 2014: 19 / 4 / 0 in 252 sessions.
• 2015: 31 / 6 / 2 in 252 sessions.
• 2016: 22 / 5 / 1 in 252 sessions.
• 2017: 4 / 0 / 0 in 251 sessions.
• 2018: 32 / 15 / 5 in 251 sessions.
• 2019: 15 / 5 / 0 in 252 sessions.
• 2020: 45 / 25 / 16 in 253 sessions.
• 2021: 21 / 5 / 0 in 252 sessions.
• 2022: 63 / 23 / 8 in 251 sessions.
• 2023: 28 / 1 / 0 in 250 sessions.
• 2024: 19 / 4 / 0 in 252 sessions.
• 2025: 29 / 7 / 3 in 250 sessions.
• 2026 (to Nov 3): 22 / 2 / 0 in 211 sessions.

2017 is the outlier in the other direction: four 1% down days in the whole year and not one 2% day. 2020 had 16 sessions down 3% or more, almost all in March, more than the 1960s and 1970s combined. 2022 had 63 one-percent down days, the third most since 1950 behind 2008 (75) and 2002 (72).

2026 so far against the record

Through November 3, 2026 the index has had 211 sessions, 22 of them down 1% or more, 2 down 2% or more and 0 down 3% or more. Scaled to a full year that is about 26 one-percent days, close to the 77-year average of 25.3 and well under the 2020s pace. The worst session of the year was 2026-06-05 at -2.64%; the best was 2026-03-31 at +2.91%. A year with two 2% down days and no 3% day is a quiet year by the standards of the last decade, whatever the headlines felt like on the day. The VIX post reads the same calm from the options market's side.

What happens the day after a 2% drop

People ask this because they want to know whether to sell into it. Across the 444 sessions since 1950 that closed down 2% or more, the next session was up 245 times, 55%, with a median move of +0.19%. Five sessions later the index was higher 256 times out of 444, median +0.61%. That is a slight lean toward recovery, not a rule. The more useful number is clustering: 175 of the 444 two-percent down days, 39%, were followed by another 2% down day within the next five sessions. Big days come in groups. A stop that is hit on the first one is often hit for a reason the second one confirms, and sometimes it is hit by a day that reverses by the close of the next.

What it means for a stop-loss

A stop-loss distance is a bet about how often you are willing to be taken out by ordinary noise. On the index alone, the base rates above say what each distance meets in a typical year, before any single stock's wider swings and before the gap between your stop price and your fill, which the stop-loss guide covers.

• A 1% stop on an index-like position would have been hit by a single session's move about 25.3 times a year, and more than 40 times a year in the 2000s. That is a stop that trades the calendar, not the thesis.
• A 2% stop meets about 5.8 such days a year on average, 9.8 in the 2020s, and four out of ten of those days bring a second one within a week.
• A 3% stop meets 1.7 a year on average and 4.0 in the 2020s; outside 2008, 2020 and 2022 most years have one or none.
• A 5% stop has been hit by a single index session 30 times in 77 years, and most of those were in the five worst months of the record. The circuit-breaker post lists every 7% day, the level at which trading halts.

Single stocks move two to three times as much as the index on the same day, which is why the same distances mean something different on a position in one name, and why the default stop and trailing stop in JorgAI sit wider than the index numbers above. A trailing stop changes the question again: it is reset by every new high, so it is measured against a pullback from the high, not a day's close-to-close move. Across the live auto trades JorgAI closed in the last week, 28 of 57 exits were a trailing stop and 10 a fixed stop, which is the system doing the exiting rather than a member deciding under a red screen. How that exit executes when the broker connection is down is its own post, and how fast the order fills once it is sent is measured on Schwab.

How we counted

Daily closing prices of the S&P 500 from Yahoo Finance, January 3, 1950 to November 3, 2026, 19,331 close-to-close moves. A day counts as a 1% down day when the close is 1% or more below the prior close; intraday lows are not used, so a day that fell 2% at noon and closed down 0.5% is not counted. Decade rates divide the count by sessions in the decade and multiply by 252. The day-after and five-day figures use closes only. Up days are counted the same way for comparison.

Quick answers

How often does the stock market drop 1% in a day?

About one session in ten. Since 1950 the S&P 500 has closed down 1% or more on 1,942 of 19,331 sessions, 25.3 times a year on average, with a range from four days in 2017 to 75 in 2008.

How often does the S&P 500 fall 2% or 3% in a day?

Down 2% or more: 444 times since 1950, about 5.8 a year. Down 3% or more: 132 times, about 1.7 a year. The 2020s run higher on both, 9.8 and 4.0 a year.

What usually happens the day after a 2% drop?

Since 1950 the next session was up 55% of the time with a median of +0.19%, and the index was higher five sessions later 58% of the time. But 39% of 2% down days were followed by another within five sessions.

How many 1% down days has the S&P 500 had in 2026?

22 through November 3, 2026, in 211 sessions, with 2 days down 2% or more and none down 3% or more. The worst session was 2026-06-05 at -2.64%.

Where should I set a stop-loss, given these numbers?

This page does not tell you a number. It tells you what each distance meets on the index in an ordinary year: a 1% stop meets ordinary noise 25 times a year, a 2% stop six, a 3% stop under two, and a single stock moves more than the index. Pick the distance that matches the move you are trying to avoid, not the one that feels safe on a red day.

Data: S&P 500 daily closes from Yahoo Finance, 1950-01-03 to 2026-11-03; counts are close to close. The JorgAI exit counts are platform-wide for one week and describe mechanisms, not results. Past frequencies are not a forecast. Education, not financial advice. Related: the best and worst months since 1950 and maximum drawdown explained.

If you want the stop to fire without you watching the screen, start a free trial and set the stop and trailing stop once; the rules do the exiting.