FWIW, I tried to create a "better by ROE" group from within the Nasdaq 100 set of stocks.
Historically I have found this is a tough group to subset: it's usually better to just buy them all and be done with it!

This is my best attempt

I did this exercise with Value Line.

I don't have an exact "Nasdaq 100" membership field, so I approximated it.
At a top level view, the Nasdaq 100 is defined as the 100 largest Nasdaq-listed stocks which are not financials.
Technically it is "top 100 HTD 105", but let's ignore that.

So, to roughly approximate the Nasdaq 100 I start off with this as a universe:
Exchange is Nasdaq
Sales > 0 (Value Line does not populate the Sales field for most financials including banks, so this is a quick and dirty way to eliminate them)
Market cap top 100

Then we add the screen steps:
Ratio of price to 52-week high, top 50 (crowd source elimination of the ones crashing most at the moment)
ROE top 30
5-year sales growth rate top 20

With friction, it looks better at two month holds.

CAGR SPY 1997 to June 2023: 8.82%
CAGR All 100 stocks of my Nasdaq crudy proxy: 9.03%/year
CAGR of this 20 stocks screen, 2 month holds with 0.4% friction: 15.13%/year

My risk metric (DDD3) of SPY: 11.28%
My risk metric (DDD3) of Nasdaq 100 proxy: 15.96%
My risk metric (DDD3) of the 20 stock screen: 10.32%

Of course, the Nasdaq 100 (proxy) stocks had such wild gyrations around the tech bubble that it's more illustrative to break it down into eras

The bubble, January 97 - March 2000
Nasdaq 100 stocks up 50%/year, this screen up 47%/year: it kept up

The crash, March 2000-October 2002
Nasdaq 100 stocks down -81%, this screen -41% (figures not annualized -- the screen offered a lot of protection, but only on a relative basis)

The bungee rebound, October 2002 - November 2003
Nasdaq 100 stocks up +88%, this screen +40% (figures not annualized -- screen did not keep up: after a crash, the worst stuff always bounces best)

More normal times, November 2003 - June 2023
Nasdaq 100 stocks up +10.9%/year, this screen +14.8%/year (nice advantage in relatively normal times, at least in backtest)
This would have beat the S&P 500 and Nasdaq 100 equal weight about 70% of rolling years in the last ~20 years. Again, in backtest.

Some recent picks, to get a feel for what it likes
MELI VRTX NFLX NVDA GOOG ADBE LULU KLAC LRCX AMAT
CPRT AAPL MSFT IDXX CDNS SNPS NXPI COST MU CTAS

Jim