Stock Correlation: Is the Market's Calm Hiding Risk?
An index is only as volatile as its stocks and the correlation between them. When stocks move on their own news, their swings cancel inside the index, and the headline reads calm while the average stock swings hard. That calm lasts only as long as the correlation stays low. This page measures how low it is, how much volatility the index would carry if correlation returned to normal, and what followed past lows. For what the options market expects correlation to be, see Volatility Premium.
Today's reading
Over the six months (126 sessions) ending October 19, 2026, the average correlation between the 500 largest US stocks was 0.073, rank 3 of 4,099 sessions since July 2010. SPY's realized volatility over the same window was 11.8%, while the average stock's was 37.8%. If correlation went back to its median of 0.26 with every stock as volatile as it has been, the index would run about 19.4%. Over one month the 100 largest stocks sit at 0.09, the 6th percentile since 2011.
Sources, methodology & freshnessOur own ~5,500-symbol daily price database (TradeStation) plus the delisted-symbol graveyard, with SPY as the index reference; Ken French’s data library (49 industry portfolios, daily CRSP returns) for the record before 2010. Market-cap ranking from the Polygon share-class snapshot, carried by each name’s split-adjusted price change. · Recomputed after every US trading close as part of the daily pipeline; the industry record follows Ken French’s monthly updates.Data as of 2026-10-19 · Open ↓Close ↑
Rule: “Calm by cancellation” when the six-month correlation is in its lowest fifth (at or below 0.179) and SPY's six-month volatility is at or below its median (13.4%). “If it reverts” is the average stock's volatility × √0.26, the median correlation.
Six-month correlation across the 500 largest stocks
The average correlation between every pair of the 500 largest US stocks over a trailing six months, the series that circulates as “S&P 500 realized correlation.” At 0.073 it is rank 3 in this record, below the Jan 2018 low of 0.088. The shaded band is the lowest tenth of the record, where the risk study in section 02 starts; earlier stretches inside it are labelled with their low, one label per year.
What followed: volatility, not direction
Low correlation is a statement about risk, so the test is about risk: did SPY's one-month volatility reach 25%, a level it hits in selloffs, and how far did the index fall? To keep the comparison fair, the table holds the index equally calm in both rows (six-month volatility at or below its median, 13.4%) and varies only the correlation.
| From each day with… | Days | Vol hit 25%, 6 mo | Vol hit 25%, 12 mo | Peak vol, 12 mo | 10%+ fall, 12 mo | Worst fall, 12 mo |
|---|---|---|---|---|---|---|
| Correlation in its lowest 10%, index quietToday | 307 | 36% | 100% | 26.8% | 28% | −1.4% |
| Index quiet, correlation not low | 1,705 | 32% | 55% | 28.9% | 39% | −5.7% |
| All days | 3,973 | 34% | 56% | 26.8% | 34% | −5.2% |
SPY, 2010 on. Vol is SPY's 21-session realized volatility, annualized; peak vol and worst fall are medians across days; worst fall is the lowest close within the year against the starting close, without dividends. Consecutive days share most of their future, so the day counts overstate the evidence: the low-correlation row is 6 separate runs, listed below.
Within six months, low correlation told you nothing. Volatility reached 25% after 36% of low-correlation days against 32% of equally calm days with normal correlation. Counted by run rather than by day, 3 of 6 runs got there within six months, and across cut-offs from the lowest 5% to the lowest 30% the share ran from 20% to 50%, against 34% of all days.
Within a year, the calm ended in every completed run. 5 of 5 runs reached 25% within twelve months, against 55% of calm days with normal correlation. Across the other cut-offs the share ran from 56% to 100%, against 56% of all days. Five runs is a thin sample, and a year is long enough that most calm stretches end in some shock.
The falls were not bigger. The median worst fall within a year was −1.4% after low-correlation days against −5.7% after calm days with normal correlation. The risk in this record arrived as volatility, not as deeper losses. Nor has the gauge carried direction: the lowest quarter of six-month readings was followed by an average SPY return of +6.57% over the next six months, against +6.59% for every month-end.
| Every run into the lowest 10% with a calm index, and the year after | |||||
|---|---|---|---|---|---|
| Run began | SPY vol then | Vol hit 25% after | Peak vol | Peak on | Worst fall |
| 2017-03-24 | 8.3% | 225 sessions | 26.8% | Mar 2018 | −0.6% |
| 2018-10-02 | 5.6% | 58 sessions | 31.1% | Jan 2019 | −19.6% |
| 2021-11-22 | 6.6% | 78 sessions | 33.7% | May 2022 | −23.7% |
| 2024-07-05 | 6.6% | 187 sessions | 51.1% | Apr 2025 | −10.5% |
| 2025-02-20 | 9.9% | 30 sessions | 51.1% | Apr 2025 | −18.7% |
| 2025-10-20still on · 250 sessions so far | 13.0% | not yet | 19.7% | Apr 2026 | −5.9% |
A run begins on the first session with the six-month correlation at or below 0.144 and SPY's six-month volatility at or below 13.4%; a new run needs 126 sessions with the condition off. Outcomes cover the 252 sessions after the start. The current run began October 20, 2025; in its first 250 sessions SPY's one-month volatility peaked at 19.7% (Apr 2026).
How much volatility the calm is hiding
For a large basket, index volatility is close to the average stock's volatility times the square root of the correlation. That makes the size of the risk computable: hold every stock as volatile as it has been and put correlation back at its median of 0.26. The dashed line is the index volatility that would follow. Today it is about 19.4%, against an actual 11.8%. The gap between the black and dashed lines is the volatility that cancellation is absorbing.
The trigger: one-month correlation
The six-month series moves slowly. When the calm ends, it shows first here, across the 100 largest stocks over 21 sessions: correlation jumps toward one in a selloff, and that jump is the moment a diversified book stops being diversified. It sits at 0.09 today, the 6th percentile since 2011, with SPY at 10.0% volatility and the average stock at 30.2%. A move above 0.42, its 80th percentile, would mean the stocks are moving together again.
The longer record: industries since 1926
Our stock-level record starts in 2010. For the decades before, the same six-month measure runs on Ken French's 49 industry portfolios, which reach back to 1926. Industries move together more than single stocks, so its level sits above the stock series in section 01, but where both exist they have moved together closely (a weekly correlation of 0.95 since 2010). Its past-year low of 0.221 (August 19, 2026) is the lowest since December 2000. French publishes monthly, so this line ends August 31, 2026.
What this gauge cannot tell you
Rule: “Calm by cancellation” when the six-month correlation is in its lowest fifth (at or below 0.179) and SPY's six-month volatility is at or below its median (13.4%). “If it reverts” is the average stock's volatility × √0.26, the median correlation.
How Stock Correlation Works
- 1Measure each stock and the basket the same wayAnnualized realized volatility from daily returns over the same window, 126 sessions for the six-month series and 21 for the one-month trigger, for every stock in the basket and for the equal-weighted basket itself.
- 2Pick the basket point-in-time, delisted names includedThe 500 largest US common stocks (and the 100 largest for the one-month series) are re-picked on the first session of every month from trailing data, drawn from the live universe AND our delisted graveyard. A company that was large in 2014 and no longer exists is in every window it actually traded through.
- 3Back out the average pairwise correlationFor an equal-weighted basket, N²·σ²(basket) = Σσᵢ² + ΣΣρᵢⱼσᵢσⱼ. Every term except the average correlation is measured, so the equation solves for it. It is Cboe’s implied-correlation construction run on realized prices instead of option quotes.
- 4Turn it into a risk figureFor a large basket, index volatility is close to the average stock’s volatility times the square root of the correlation. Holding stock volatility fixed and putting correlation back at its median gives the index volatility the calm is hiding. The page also tests what followed every past low in volatility and drawdown terms, against equally calm days with normal correlation.