But none of this contradicts the fact that the biggest companies don't USUALLY do better than the average.
...
I don't get it. How did the biggest companies become the biggest if they weren't better than average? Seems to me that they became biggest BECAUSE they were better than average.

Think of it this way:
Being the best business over a long time will get you to being one of the biggest in terms of true intrinsic value, say into the top 30.
But the easiest way to get that boost to the top 5 by market cap within that group is to become overvalued for a while.
Consequently, the odds that one of the top 5 is currently overvalued is considerably higher than the odds for any other size of firm.

If you accept the possibility that companies *can* be overvalued and undervalued, that's pretty much a mathematical certainty. Create any sequence of numbers with 1/n (Zipf's Law) distribution to represent true IV sizes of a hypothetical set of companies, add or subtract (say) 40% from each one randomly, sort, and see how much the top 5 had been boosted by the random addition. At any other size range the average of several consecutive over/under percentages will tend towards zero, but not at the top where they will tend to be large. (and at the very bottom where they will tend to be small, but nobody pays attention down there)

The only exception is if the top 5 firms have true intrinsic values that are so much higher than all the others in the top slice that even if every firm is +/- some big random number, the top 5 don't change to other ranks when sorted. It is possible that we are seeing a bit of that in recent years. Or not.

Without a doubt, the recent result is very good for the biggest, and also without a doubt that is unlike deeper history. Compare the top N by market cap to the S&P 500 (a fair part of which is the same 5!)

1986 - 1992  (7.0 years) Top 5 underperformed by -3.7%/year, top 4 by  -4.8%
1993 - 1999  (7.0 years) Top 5 underperformed by -3.5%/year, top 4 by  -2.7%
2000 - 2006  (7.0 years) Top 5 underperformed by -5.7%/year, top 4 by  -6.8%
2007 - 2013  (7.0 years) Top 5 underperformed by -4.7%/year, top 4 by  -4.8%
...but then...
2014 - 2018  (5.0 years) Top 5  outperformed by  +5.8%/year, top 4 by  +8.7% (breakdown of the usual rule)
2019 - 2024  (5.4 years) Top 5  outperformed by  +9.8%/year, top 4 by +14.2% (HUGE breakdown of the usual rule)

So, the recent outperformance of the very largest is some currently unknown mix of (a) a breakdown in the Zipf Law type distribution of company sizes due to things like some winner-takes-all effects, and/or (b) gigacap valuation bubble like the late 1990s, and/or (c) pure coincidence that these firms just happen do be doing well, for reasons unrelated to their size. It's not all (b), that's for sure, since the very largest firms are, with a couple of exceptions, extremely profitable and cash generative, so it's not as bubbly in that sense compared to the tech bubble.

Jim