What a Market Correction Means by the Numbers

A market correction is a decline of 10 percent or more in an index from its most recent peak, usually measured on closing prices. Since 1950, the S&P 500 has averaged an intra-year drawdown of 13.6 percent, per A Wealth of Common Sense's January 2022 analysis of index data.

HORISON publishes information, not investment advice. This explainer covers what the documented threshold means and what history shows; it does not predict market direction, and allocation decisions depend on individual circumstances this site cannot know.

What counts as a correction, exactly?

The working definition is arithmetic, not official. Analysts conventionally label a decline of 5 to 10 percent a pullback, a decline of 10 to 20 percent a correction, and a decline of more than 20 percent a bear market. No regulator or index provider enforces these bands; they are shorthand that financial commentary adopted over decades.

Three measurement choices change what gets counted. The decline is typically computed from the highest closing level of the index to the lowest subsequent close, which filters out single-session intraday spikes. The definition is index-specific — the Nasdaq can correct while the S&P 500 does not. And the peak reference is always the most recent one, so a market that falls 9 percent, recovers, and falls 9 percent again has had two pullbacks and no correction.

A worked example makes the arithmetic concrete. An index that closes at a peak of 5,000 enters correction territory below 4,500 (−10 percent) and bear-market territory below 4,000 (−20 percent). If it bottoms at 4,620, the episode is a 7.6 percent pullback; if it bottoms at 4,410, it is an 11.8 percent correction, even though the two outcomes differ by only 210 index points.

How often do corrections happen?

More often than most investors expect. An analysis of S&P 500 data published by A Wealth of Common Sense in January 2022 found that a correction of 10 percent or more has occurred roughly once every two years on average since 1950, and a bear market roughly once every seven years. The same analysis put the average intra-year drawdown at 13.6 percent — meaning the typical calendar year contains a decline larger than the correction threshold itself.

Smaller dips are even more common. Per Invesco analysis of S&P 500 history since the early 1980s, the index has seen a drawdown greater than 5 percent in every year but two. The practical implication is that drawdowns of some size are a background condition of holding equities, not a departure from normal functioning.

What have documented corrections looked like?

Four episodes since 2018 illustrate the range. Each figure below is computed from S&P 500 daily closing levels, per S&P Dow Jones Indices historical data, with the peak and trough dates shown.

Two observations follow. First, speed varies widely: the 2020 episode covered 33 trading days from peak to trough, while the 2022 decline unfolded over more than nine months. Second, the 2018 and 2025 episodes stopped just short of the 20 percent bear-market line, which shows how the same approximate decline can carry different labels depending on where it stops.

How does a correction differ from a crash?

The two words describe different dimensions. A correction is about depth — a percentage distance from a peak, reached over any number of sessions. A crash is about speed: an abrupt single-session or multi-session collapse. The most documented example remains October 19, 1987, when the Dow Jones Industrial Average fell 22.6 percent in one trading day, per Dow Jones Indexes history — a decline deep enough for a bear market label, delivered in a single session rather than months.

Because the correction label is depth-based, an index can correct without anything that commentators would call a crash, and a crash can occur from a level well below a recent peak. The categories answer different questions — how far and how fast — and commentary that mixes them says less than either word alone.

How long did recoveries take after recent episodes?

Recovery times varied widely across the documented episodes, measured from trough close to the first new closing high, per S&P 500 closing history from S&P Dow Jones Indices. After the December 24, 2018 trough, the index reached a new closing high on April 23, 2019 — about four months. After the March 23, 2020 trough, a new closing high arrived on August 18, 2020 — under five months.

The 2022 bear market took materially longer. From the October 12, 2022 trough, the S&P 500 did not print a new closing high until January 19, 2024, roughly fifteen months later. The spread — four months to fifteen — is itself the lesson: recovery duration is not a stable constant that can be planned around, and averages across episodes smooth over episodes that felt very different to live through.

What does the 10 percent threshold actually signal?

By itself, nothing mechanical. The threshold does not trigger any market mechanism — unlike circuit breakers, which halt trading at defined intraday declines. Crossing minus 10 percent does not force selling, change margin requirements, or alter index construction. The label is descriptive, and its main effect is on communication: media coverage tends to intensify once the word correction applies.

The threshold also carries no documented predictive content. A 10 percent decline has historically resolved into a deeper bear market in some episodes and reversed into new highs in others, and no published rule separates the two in advance. Treating the label as a signal confuses description with forecasting — the site presents the categories so readers can weigh commentary that uses them.

How do corrections interact with portfolio strategy?

Framework responses treat a correction as expected variance rather than new information. A written allocation with a rebalancing rule responds mechanically: falling equity prices shift weights away from target, and rebalancing moves them back on the schedule the investor set in advance. The documented trade-off is discipline versus tax and transaction costs in taxable accounts.

Two further points qualify the picture. Diversification dampens but does not remove drawdowns — in 2022, both stocks and intermediate Treasuries fell, leaving balanced portfolios with losses smaller than equities alone but not zero. And a correction in one asset class says nothing about others: crypto assets have historically experienced drawdowns several times deeper than equity corrections in the same periods, which is why they are treated here inside the same risk framework rather than a separate one. Crypto assets can lose most or all of their value quickly.

What should readers take from the label?

The documented record supports a narrow set of conclusions. Corrections are frequent, thresholds are conventions rather than mechanisms, and historical episodes differ widely in depth, speed, and what followed. What an investor does about that depends on time horizon, obligations, and risk capacity — individual circumstances that no general reference piece can know.