Cointegration vs. Correlation

Cointegration is a stationarity of the spread between two instruments or variables whereas correlation is when the returns or results of two instruments or variables move in the same direction and is often associated either the short term or long term properties of a time series. Cointegration is present if a relationship among two non-stationary series exists where the residuals of the regression are stationary.  or    is stationary when x and y confer cointegration. Alternatively, the correlation coefficient of two variables or instruments x and y is defined as r=Cov(X,Y)/std(X).std(Y) where Cov is covariance and std is standard deviation.