Hi all,

I have a data series of stock returns that I'm trying to run regression on. I'm looking to test for auto-correlation using the Breusch Godfrey test but I'm not exactly sure of how I determine the number of lags I should use in the test?