What stock correlation measures
Pearson correlation summarizes linear co-movement between paired return observations. It ranges from −1 to +1 when both series have nonzero variance. Zero does not imply independence, and a correlation of one does not mean equal returns or volatility.
How to use the matrix
- Add two to ten supported symbols and choose a requested calendar window.
- Calculate, then inspect the actual common start date, end date and number of aligned return intervals.
- Compare pairs across windows. Adding a symbol can shrink the common sample and change every cell.
- Check original prices and corporate actions before using an unusual result.
Alignment and formula
The calculator requests FMP historical-price-light data, excludes invalid or duplicate observations, and intersects available dates across all symbols. It requires at least 30 common prices, yielding at least 29 return intervals. Return equals current price divided by previous common-date price, minus one. Missing sessions can make an interval longer than one trading day; prices are not forward filled.
Correlation is the sum of paired demeaned return products divided by the product of the two demeaned sum-of-squares square roots. A constant-return series has an undefined correlation; the tool reports an error rather than calling it zero. The displayed diagonal is one only after valid pair calculations.
Read colors as display bands
The heatmap uses ±0.3 and ±0.7 as visual cutoffs. These are not statistically tested trading thresholds. The summary averages distinct off-diagonal pairs without portfolio weights. It does not calculate portfolio volatility, hedging size or an optimal allocation.
Price returns and practical limits
This calculation uses the provider price field, with no separate dividend reinvestment or currency conversion. Different trading hours, price adjustments and exchange calendars can affect comparison. A longer history can mix regimes, and past correlations may fail during stress. Evaluate weights, individual volatility, liquidity and adverse scenarios separately.
For a mathematical reference, see NIST: correlation. Provider coverage is described in FMP historical end-of-day light data. No future return or diversification benefit is guaranteed.