quant screening on a universe of stocks
 

Are uni and/or multivariate factor screens useful to filter for possible investment candidates? And if so, can someone point to good research on the subject? i.e. scanning and ranking the entire S&P bank sector by factors such as earnings growth, P/B ratio, and a few other financials, and determining that the top 20 should be considered as long candidates and the bottom 20 short candidates.

I have seen research on the matter but it was not convincing. Nonetheless, an initial quant screening might save a lot of time to focus in on outliers that deserve special attention.