Calculate a single "best fit" multivariate linear regression, optimized across multiple dependent variables concurrently

I have various valuation metrics (P/B, P/E, historical returns, etc.) and I'm trying to predict forward-looking returns over multiple time periods (1-5 years). However, I don't want to calculate a separate model for each time-period. Instead, I want to calculate a single, "best fit" model that maximizes the fit across each of the time periods at the same time. Any ideas?

[Brendan Hamm]

You could do a multivariate regression using mvregress. This is what it sounds like you are trying to do. I wouldn't recommend trying to extrapolate future values from a linear regression though. Other considerations include neural networks ( narxnet or narnet in particular), or VAR ( vgxset).