Give me some wisdom.

Ha! You'll tend to get a different answer from different people. But going way out on a limb, I think that most here would say that it has been extraordinarily difficult to use MI to beat an index for quite a number of years now.

Partly because the index has done so well, making all timing and individual stock selection seem like a waste of time, but many have concluded that the simple factors that used to work don't seem to work any more. They never worked perfectly, but some of them seemed to work for a very long time.

A classic example is YLDEARNYEAR. It was published in 2003. It beat the S&P 500 by 10%/year (monthly top 10) in the next ten year 2004-2013, which seemed at the time to be definitive proof that some of the MI screens were far more than just overmining the data. But then it trailed the index for the next six years by 12%/year (admittedly while ALL dividend stocks were lagging a lot on average), then fell off a cliff, hit by the 2020 pandemic which created the mother of all whipsaws for anything momentum based--lagging the index by 42% that year. It then went on to beat the S&P by 15%/year 2021-2024.
So...is it a good screen or not?

I have had modest success with very diversified screens with very modest goals. e.g. LargeCapCash, a 40 stock screen aiming to beat the market by maybe 2-3%/year with very low risk. It beat the S&P by 4.7%/year in the first 4.75 years after publication, using "top 40 HTD 45", trading each two months, with friction. The variant that requires a dividend and a valid Timeliness rank. 4.4% if you didn't require a valid Timeliness.

Jim