When we run a regression we assuming that the independant variable explains the dependant variable. So for a linear or log-linear trend model are we saying that TIME explains out dependant variable? Also for AR models are we saying the variable itselfs (past instances) explains the current instance? What is the practical value of that?..

forecasting and performance attribution

forecasting I can see, but how would performance attribution work? Say for a trend model. Would it be excess returns above the model must be attributed to something else besides the trend?

mwvt9 Wrote: ------------------------------------------------------- > forecasting I can see, but how would performance > attribution work? > > Say for a trend model. Would it be excess returns > above the model must be attributed to something > else besides the trend? performance attribution is more appropriate for regression models.