Direct Answer

Correlation for technical analysts is a statistical measure, typically the Pearson correlation coefficient, of how closely two assets' price movements track each other over a chosen lookback period. It is expressed on a scale from -1 (moving in perfect opposite directions) to +1 (moving in perfect lockstep), with 0 indicating no consistent linear relationship. Traders use it to spot relative-strength relationships, evaluate real diversification, and identify candidate pairs for spread-based strategies.

Key Takeaways

  • Correlation ranges from -1 (perfect inverse relationship) to +1 (perfect positive relationship), with 0 meaning no linear relationship.
  • The Pearson correlation coefficient is the standard measure, calculated from the returns (not raw prices) of two assets over a shared period.
  • Correlation is not causation, two assets can move together because of a shared external driver rather than one influencing the other.
  • Correlation is not static: the relationship between two assets can strengthen, weaken, or flip sign as market conditions change.
  • Technical analysts use correlation to check whether a portfolio is genuinely diversified, since holding many assets that are all highly correlated offers less risk reduction than it appears.
  • Pairs trading strategies typically screen for historically high correlation, then look for temporary divergence between the two assets as a trade signal.
  • Rolling correlation (recalculated over a moving window) is often more useful than a single static figure, since it reveals whether a relationship is stable or drifting.
  • Common conventions treat |r| above about 0.7 as strong, 0.3-0.7 as moderate, and below 0.3 as weak, but these are informal thresholds, not fixed rules.

What Is Correlation in Technical Analysis?

Correlation measures the strength and direction of the linear relationship between the price movements of two assets. Rather than comparing raw price levels, which can be misleading, since two unrelated assets can both trend upward over time purely by coincidence, technical analysts typically calculate correlation from period-over-period returns. This isolates whether the assets tend to rise and fall together, move in opposite directions, or show no consistent pattern relative to each other.

The most widely used measure is the Pearson correlation coefficient, denoted r. A value of +1 means the two assets' returns moved in perfect lockstep over the sample period; a value of -1 means they moved in perfectly opposite directions; a value near 0 means there was no reliable linear relationship between them at all.

How the Correlation Coefficient Is Calculated

The Pearson correlation coefficient between two return series, X and Y, is defined as:

r = Cov(X, Y) / (σX × σY)

Where Cov(X, Y) is the covariance of the two return series, a measure of how they vary together, and σX and σY are the standard deviations of each individual series. Dividing covariance by the product of the two standard deviations rescales the result to always fall between -1 and +1, which is what makes the coefficient comparable across any pair of assets regardless of their individual volatility.

In practice, traders rarely compute this by hand. Spreadsheet functions (such as CORREL) and charting-platform indicators calculate rolling correlation automatically once a lookback window, commonly 20, 60, or 90 trading days, is specified.

Worked Example (Hypothetical)

Consider a hypothetical scenario comparing the daily returns of two fictional stocks, "Stock A" and "Stock B," over a 10-day illustrative window. Suppose that on most days when Stock A rose 1%, Stock B rose roughly 0.8%, and on days when Stock A fell, Stock B tended to fall by a similar proportional amount, with only minor day-to-day deviations from that pattern. Running the Pearson formula on this hypothetical return series might produce a coefficient of approximately r = 0.85, indicating a strong positive relationship.

By contrast, imagine a second hypothetical pair, "Stock C" and "Stock D", where Stock C's up days were about as likely to coincide with Stock D's down days as with its up days, and the two showed no consistent proportional pattern. That hypothetical pair might produce a coefficient near r = 0.05, indicating essentially no linear relationship. These figures are illustrative only and do not reflect any real securities.

Why Correlation Matters for Traders

Correlation gives technical analysts a way to quantify relationships that a price chart alone can only suggest. A portfolio holding ten stocks that are all highly correlated with each other behaves, in risk terms, much closer to holding one large position than ten independent ones, correlation analysis is what reveals that concentration, which position count alone does not show. Traders building or rebalancing a portfolio often check pairwise correlation to confirm that adding a new position is actually reducing, rather than merely disguising, overall risk.

Correlation is also foundational to pairs and statistical-arbitrage trading, where a strategy identifies two historically correlated assets and trades the spread between them when that relationship temporarily stretches. And on a simpler level, tracking an individual stock's correlation to a broader index or sector can help a trader judge how much of that stock's move is idiosyncratic versus driven by the wider market or sector trend.

Limitations and Common Mistakes

  • Confusing correlation with causation. A high correlation coefficient shows association, not that one asset's price is driving the other's, both may simply be responding to a shared factor like broad market sentiment.
  • Treating correlation as fixed. Correlation is measured over a specific historical window and can shift meaningfully in a new market regime; a relationship that held for months can break down without warning.
  • Using price levels instead of returns. Comparing raw price series rather than period returns can produce a spuriously high correlation between two assets that simply both trended in the same broad direction over time.
  • Ignoring correlation instability during stress periods. Many previously uncorrelated or weakly correlated assets tend to move toward higher positive correlation during broad market sell-offs, reducing the diversification benefit exactly when it is needed most.
  • Over-relying on a single static figure. A correlation coefficient calculated once over a long historical period can mask meaningful short-term drift; rolling correlation over a moving window gives a more current read.
  • Assuming correlation implies a tradeable relationship. A high historical correlation between two assets does not guarantee the relationship will persist long enough, or predictably enough, to support a profitable pairs trade.

The Number That Moves When It Matters Most

The most consequential property of a correlation coefficient is that it is not stable, and the instability is not random. Assets that show weak or negative correlation through calm conditions frequently move toward strong positive correlation during broad sell-offs, which means the diversification a portfolio appeared to have tends to decline exactly when it was supposed to help. A correlation matrix built from a quiet period describes a quiet period.

The calculation itself has a common technical error worth avoiding. Correlation should be computed from period returns, not from raw price levels. Two assets that both trended upward over the same years will show a high coefficient from their price series regardless of whether their day-to-day movements have anything to do with each other, and that number is an artefact of shared drift.

Then there is the interpretation gap. A high coefficient reports association and says nothing about mechanism, and the most common real explanation is that both assets are responding to a third factor rather than to each other. For pairs-based approaches, that distinction matters a great deal: a relationship maintained by a shared driver can end when the driver changes, without anything visible happening to either asset.

Practically, treat any coefficient as belonging to its window and re-measure it across different market conditions. A relationship that held for months can break without warning, and the break tends to show up in your positions before it shows up in your statistics.

Frequently Asked Questions

What is correlation in technical analysis?

Correlation in technical analysis is a statistical measure, typically the Pearson correlation coefficient, of how closely the price movements of two assets track each other over a given period. It ranges from -1 (perfect opposite movement) to +1 (perfect same-direction movement), with 0 meaning no linear relationship.

How is the correlation coefficient calculated?

The Pearson correlation coefficient divides the covariance of two assets' returns by the product of their individual standard deviations. In practice, traders compute period-over-period returns for both assets over a lookback window, then apply this formula, often using spreadsheet or platform functions rather than calculating by hand.

What correlation value is considered strong?

There is no universal cutoff, but many traders treat readings above roughly 0.7 or below roughly -0.7 as strong, values between about 0.3 and 0.7 (or -0.3 to -0.7) as moderate, and values near zero as weak or negligible. These thresholds are conventions, not statistical laws, and should be interpreted in context.

Does high correlation mean one asset causes the other to move?

No. Correlation measures association, not causation. Two assets can move together because of a shared underlying driver, such as broad market sentiment or sector-wide news, without either one causing the other's price action.

How do traders use correlation for pairs trading?

Pairs trading strategies typically look for two historically highly correlated assets whose prices temporarily diverge, then take a long position in the relatively underperforming asset and a short position in the relatively outperforming one, anticipating the relationship will revert. This depends on the correlation remaining stable, which is not guaranteed.

Should correlation be computed on prices or on returns?

On returns. Two price series that both trend upward will show a high correlation coefficient regardless of whether their day to day movements are related, because the shared trend dominates the calculation. Differencing to returns removes the trend and measures whether the moves themselves co-occur, which is the question almost everyone is actually asking when they compute the number.

What is rank correlation and when is it preferable?

Rank correlation replaces each observation with its position in the ordering and correlates those ranks. It is less affected by a single extreme observation, which matters in return data where one session can dominate a Pearson calculation, and it captures relationships that are consistent in direction without being straight-line. It answers a slightly narrower question in exchange for being harder to distort.

How does window length change a rolling correlation?

A short window produces a much noisier series that reaches values close to plus or minus one regularly, simply because few observations are involved and a handful of aligned moves is enough. A long window is steadier and slower to register a genuine change in the relationship. Neither is correct, and a rolling correlation chart is not interpretable without knowing which was used.

Does a correlation near zero mean two assets are unrelated?

It means no linear relationship was detected over that window, which is a narrower claim. Two series can be strongly dependent in a way the coefficient cannot see, for example moving together only during large moves and independently otherwise, and produce a coefficient near zero. That pattern matters for exactly the periods when the relationship is most consequential.

References

Disclaimer

This page is for educational purposes only and does not constitute investment, financial, or trading advice. Correlation is a backward-looking statistical measure and does not guarantee that a historical relationship between two assets will continue. Any example, chart, or figure on this page uses illustrative, hypothetical data rather than live market data. Swoopr Investment is not a licensed investment advisor; consult a qualified professional before making investment decisions.