Correlation（资产相关性）
A sample-dependent measure of linear co-movement—not proof of independence, causation, or stable diversification.
Correlation is a standardized measure of linear co-movement between two variables, usually bounded from -1 to +1 when defined. A value near zero indicates little estimated linear association in the selected sample, not necessarily no relationship, independence, or no shared risk.
Frequently Asked Questions
Does zero correlation mean two assets are unrelated?
No. They may have nonlinear dependence, common tail exposure, changing lead-lag relationships, or correlations that offset over the sample. Zero estimated linear correlation is not proof of independence.
Are historical correlations stable?
No. Estimates depend on frequency, window, currency, benchmark, data quality, and market regime. Correlations can change abruptly and often rise during stress.
Can low correlation guarantee lower portfolio losses?
No. Portfolio outcomes also depend on weights, volatility, liquidity, leverage, tail dependence, valuation, and whether correlations change when losses occur.
Should price levels or returns be correlated?
For portfolio analysis, returns are usually more appropriate. Correlating trending price levels can produce misleading results, while return definitions, intervals, and missing data still matter.