Subject: Re: Combs to leave
Finding 100 that should underperform by statistical analyse seem very possible for me. I'm surprise nobody tried to market such a fund. Jim did you test an approach like that?

Good news: yes.
Bad news: yes, thousands of such approaches : )
99% of which are spurious historical coincidences, I assume.

Best one-check filter is ROE, as far as I can tell.
As a random example, 1986-2024 (a pretty good sample of market regimes), an equally weighted portfolio of the top 30% of the Value Line 1700 group beat the S&P 500 by +2.18%/year. Or +1.60% if hunting only among the top 150 by market cap. I've always felt that surely there is some way to capture a bit of that tail wind...

Jim