Is this regression useful or junk
This summer instead of taking classes I've used a good majority of my time doing my own things and continued to build on my website. I started working on building regression analysis in SAS using a dependent variable of either P/TBV, P/FCF or EV/EBITDA and trying to predict those by way of operating margin, assets/liabilities, estimated growth rates, etc... With varying degrees of applicability/success (doing this by sector for the S&P500 S&P400). Typically looking at the ones with the most applicable discount to predicted (valuation metric above). All minding that the model is decent-(ly) capable statistically (Adjusted R squares, p-values of metrics are useful, Cooks-D, checking for heteroskedastic data, GOF. After companies have clearned this I read their 10-K's/10-Q's and model those companies out as the regression is just a screener.
Question: Is this regression even worth my time? Is it a useful skill to have? Any parting advice? I've noticed some sectors fit the regression better than others. And yes I have built an excel sheet to data mine all of the data I need for all (900~) companies (regressions are run daily). I am reading a 10-Q/K a day and modeling out 2-3 companies a week.
Thanks