Period: 2015-12-01 to 2025-01-31
Initial capital: $100,000
Best strategy: Top 3 / HTD 10 @ 31.8% CAGR

Some comments:
* Sortino Ratio is a better metric than Sharpe Ratio. But everybody seems to post Sharpe because ... who is this Sortino guy?

* The volatility of the holding numbers below 4 are just too high to stomach. Also the MaxDD.

* The purpose of a HTD is to reduce the trading turnover of stocks that just fell slightly beyond the cutoff. So it should be a little bit more that the N. 1HTD10 is too big. Better would be 1HTD2. 3HTD10 is too much, better would be 3HTD5.

Just personally, I would never do a top 3 or 5. That's just too likely to have a bad outcome that would take you out of the game.
If one stock craters, that's a huge chunk of the portfolio. Better losing 1 of 10 than 1 of 3.
Of course that works the same way on the upside, but missing out a double of 1 of 3 stocks doesn't take you out of the game.

Maybe doable for a "trial portfolio" with just a small amount of money that would be okay to lose it all.
But even 100% return with a small amount of money is still just a small amount of money, so why bother?

I have run backtests on various sizes. Top 5, 10, 15, 25, 50. You kind of have to eyeball how fast the CAGR drops off as the N gets larger, and how fast the volatility and MaxDD increase and pick your sweet spot.

You have to be _very_ careful when comparing CAGRs, especially considering dividends. I ran QQQ and SPY in testfol.io for those date and got:

QQQ (no div)        18.8%
QQQ                 19.7%
SPY (no div)        12.8%
SPY                 14.8%

Your table showed QQQ 18.8% and SPY 14.1%. So it looks like QQQ (no div) and SPY (incl div). So not a like-to-like comparison.

Anyway...running NAs100 (annual rebalance) in gtr1 for those dates shows:
Top 100: 14.8%
Top 50: 16.6%
Top 25: 18.1%
Top 15: 19.2%
Top 10: 19.5%
Top 5: 24.5%
Volatilities: 15%, 17%, 20%, 26%, 32%, and 44%.
BTW, you have to rebalance occasionally otherwise it will just be B&H of the very first picks.

Ooh, looks like for this date range you've have been best just staying in QQQ.

had a hard time finding Nasdaq 100 constituent stocks going back before 2014.
Ha! You should have see us in around 1999 when TimberFool and I were trying to gather the historical DJIA constituent stocks. I spent hours in the Schaumburg Library reading old newspapers on microfiche.