What Does Correlation to SPX Mean?
Correlation to SPX measures how closely a strategy's returns move with the S&P 500, on a scale from −1 to +1. Squaring it gives the share of variance explained by the index: a correlation of 0.8 means 64% of the strategy's movement is the market, not the trader.
How should you read the number?
Correlation r runs from −1 to +1. Squaring it gives r², the fraction of variance the index explains:
| Correlation | r² (variance explained) | Reading |
|---|---|---|
| 0.95 | 90% | An index fund with extra steps |
| 0.80 | 64% | Mostly market exposure |
| 0.50 | 25% | Meaningfully independent |
| 0.20 | 4% | Largely its own thing |
| −0.40 | 16% | Moves against the index |
The r² column is the one that changes minds. A correlation of 0.8 sounds like partial independence and is in fact 64% market.
Why does it matter for evaluating a record?
Because a strategy that is 90% correlated to SPX in a year the index rose 24% has not demonstrated much. The relevant comparison is not "did it make money" but "did it make more than the exposure it carried", and the correlation tells you how much of that exposure there was.
It also matters for combining strategies. Two uncorrelated positive-expectancy strategies reduce portfolio volatility together; two strategies at 0.9 correlation are one strategy in two accounts, and sizing them as if they were independent doubles the real risk.
How is correlation different from beta?
Correlation measures the strength of the relationship; beta measures its magnitude. A strategy can have a correlation of 0.9 and a beta of 0.3 — it moves with the market almost every day, but only a third as far. Or a correlation of 0.4 and a beta of 1.8, moving unpredictably but violently when it does move.
Beta = correlation × (strategy volatility ÷ index volatility). Both are worth having; neither substitutes for the other.
What can it not tell you?
Regime changes. Correlations rise in crises. A strategy uncorrelated in calm markets frequently becomes correlated exactly when that matters most.
Non-linear exposure. Options strategies have correlations that change with price, so a single number over a period is an average of very different states.
Short windows. Correlation over 30 days is a noisy estimate. Reading it over several windows is more informative than reading one.
How does this connect to a track record?
The reason a return has to be judged against something is the premise of every risk-adjusted measure. Sharpe's own 1994 restatement of his ratio defines it over the return differential against a benchmark rather than over raw return[1] — correlation to SPX is the prior question: how much of the differential is the benchmark showing up again under another name.
A correlation is an estimate with a standard error like any other, and Lo's work on Sharpe ratios makes the general point sharply: a performance statistic quoted without its sample size is not yet a claim.[2] Harvey, Liu and Zhu's argument that finance should demand a t-statistic above 3.0 applies directly to anyone claiming their returns are uncorrelated with the market.[3]
kappi publishes correlation to SPX over three windows rather than one, because a single window hides exactly the regime dependence that matters. A metric is a summary of a record, so it inherits every weakness of that record. Computed from trades selected after the fact, it is a number about the selection. kappi commits each trade before it resolves and publishes it on a Merkle-anchored log, so PnL, RME, correlation to SPX, mean R:R and trade count over 30, 100 and 200-day windows are computed over everything that was committed, losses included. $15/month to keep a record; reading one is free.
Sources
- Sharpe, 'The Sharpe Ratio', Journal of Portfolio Management 21(1), 1994, 49–58 read 2026-08-16
- Lo, 'The Statistics of Sharpe Ratios', Financial Analysts Journal 58(4), 2002, 36–52 read 2026-08-16
- Harvey, Liu & Zhu, '… and the Cross-Section of Expected Returns', Review of Financial Studies 29(1), 2016, 5–68 — argues a newly claimed factor should clear a t-statistic above 3.0 read 2026-08-16
Frequently asked questions
What does a correlation of 0.8 to the S&P 500 mean?
That 64% of the strategy's variance is explained by the index, since r² = 0.64. It sounds like partial independence and is mostly market exposure.
Is low correlation to SPX good?
It is informative rather than good. Low correlation means the returns are not simply market exposure, which makes a positive expectancy more interesting and makes the strategy more useful in a portfolio.
What is the difference between correlation and beta?
Correlation measures how consistently two things move together; beta measures how far. A strategy can have 0.9 correlation and 0.3 beta — moving with the market daily but only a third as much.
Why measure correlation over several windows?
Because correlations change with regime and rise in crises. A single window averages very different states, which is why 30, 100 and 200-day figures are more informative than one number.