I want to describe the direction of some stock returns, using as predictors several independent variables which are uncorrelated. The relation in which I am interested is between the stock returns and an independent variable which is autocorrelated. To describe the direction I use a binary logistic regression, with a binary independent variable that takes values: 0 if the stock return is negative and 1 if the stock return is positive. My question is: what I obtain if I bootstrap a binary logistic regression, in that situation? It could be correct or it does not make any sense? And why?