February 4th, 2011, 5:56 pm
I am running a simple pairs trading model based on equity correlation. I am looking to explore a more refined filter for determining whether to trade a pair than my current minimum R squared value filter. From someone with experience, what is the best avenue to begin this approach? I have read that the Johansen test for cointegration is better to use that ADF, and have considered VECM models as well. I am very familiar with Linear Regression and I understand the concept of stationarity and cointegration but have never implemented these numerical processes in excel before. I'd like some guidance going from my time series of asset prices to the output of whether they are cointegrated order I or not.