In layman’s terms, cointegration is a measure of how time series data representations are related to each other. The notion is that the difference in behavior of time series graphs can be attributed to a mathematical representation that captures their difference into a variable, P. Whenever that difference is fully aligned (P=0) the time series are considered to be acting perfectly based on historical correlation. In other words, per expectations, hence, co-integrated.

The same concept can be applied to highly correlated securities. When a pair of securities move predictably in relationship to each other, and that tight correlation is broken, it is fairly safe to assume that eventually they will become correlated again.

For the full article, visit: Utilizing Cointegration for FRX by Erez Katz