Beta vs Correlation: Two Numbers People Keep Confusing
One says how closely things move together. The other says how far. Measured values for real securities show why the difference matters.
Both describe how a security relates to the market and they answer different questions. Using one where the other belongs is the most common analytical error in portfolio discussion, and it is easy to demonstrate rather than assert.
The definitions
Correlation measures how consistently two things move in the same direction, on a scale from −1 to +1. It says nothing about magnitude. A stock that reliably moves 0.1% every time the market moves 1% has a correlation near 1.0.
Beta measures how far something moves for a given market move. A beta of 2 means a 1% market move corresponds, on average, to a 2% move in the security.
They are connected by one equation:
beta = correlation × (volatility of the security ÷ volatility of the market)
So beta is correlation scaled by relative volatility. Two securities with identical correlation can have wildly different betas, and two with identical betas can have very different correlations.
Seen in real numbers
| Ticker | Volatility | Beta vs SPY | Correlation vs SPY |
|---|---|---|---|
| NVDAHigh both | 38.0% | 1.91 | 0.648 |
| GLDLow correlation, mid beta | 29.2% | 0.76 | 0.332 |
| TLTLow both | 9.1% | 0.17 | 0.242 |
| XLFMid correlation, mid beta | 14.6% | 0.62 | 0.543 |
| KONegative, this window | 18.7% | -0.26 | -0.175 |
| QQQVery high correlation | 19.6% | 1.42 | 0.929 |
Measured against SPY from daily returns over the past year.
Read those rows against each other. Gold's beta of 0.76 looks like a moderately market-sensitive holding — until you see its correlation of 0.332. The beta is not describing a dependable relationship; it is a low correlation multiplied by a large volatility ratio, because gold is roughly 2.3 times as volatile as the index.
That is the whole point. A beta computed from a weak relationship is arithmetically valid and practically meaningless — it tells you about an average that individual days rarely resemble.
R-squared, the number that should accompany beta
Correlation squared is the share of a security's movement explained by the market. QQQ's correlation of 0.929 means roughly 86% of its variation is market-driven — its beta is a reliable description. Gold's implies only about 11%, so nearly all of gold's movement is about gold.
A beta quoted without its R-squared is half a statistic. Fund factsheets that print beta alone are inviting exactly this misreading.
Which to use when
- Diversification — correlation. You are asking whether things move together, and magnitude is irrelevant to that question.
- Finding a substitute — correlation, plus a check that the volatilities are comparable. High correlation at half the volatility is a different position, not a replacement. This is what the correlation finder shows in two columns.
- Sizing market exposure — beta. If you want a portfolio to behave like 80% of the index, beta is the number you are managing.
- Hedging — beta for the hedge ratio, correlation to judge whether the hedge will actually work. A high-beta, low-correlation hedge is a coin flip at scale.
Both are unstable, and they break together
These are measurements over a window, not properties of a security. KO's negative reading above is a real measurement over a real year and it will not hold indefinitely — defensive staples are not usually negatively correlated with equities.
Correlations also rise across risk assets during genuine market stress, which means beta rises with them. The diversification you measured in calm conditions is smaller than it looked precisely when it is being called on. Comparing a one-year figure against a three-year one, as the correlation pages do, is a cheap way to see whether a relationship is structural or an artefact of the window.
Figures on this page are measured from daily returns over the year to 2026-09-10, across 5,250 liquid US securities, and are rebuilt nightly. Reference material, not investment advice.