I am working with the US CRSP universe and have constructed a time series of monthly price-to-earnings back to the early 1960s. I have this for both individual securities as well as for portfolios sorted by certain characteristics (mostly just quintile portfolios of different exposures or metrics).
For some quintile portfolios that are associated with high risk securities, I see sometimes erratic and very negative price-to-earnings, and it's hard to know when/where I should focus debugging efforts.
What are some sources that might provide plausible bounds for such a time series. I know there are some convenient sources, like multipl.com, for quick checks of overall market P/E. But the same bounds might not matter much for risky quintile portfolios, for example. I'm not sure that a high risk portfolio will have bounds that resemble the market's, even if it does contain 20% of the cap.