Your conclusion that stats are not a good judge might be true. But tour reasons are probably wrong. It's simpler than that.
Possession reduces opportunity:
If you have the ball 65% of the time you simply have fewer opportunities to showcase your tackling and intercepting skills. Compared to someone who's team gas the ball 35% of the time.
That said, you can take teams at the upper end of possession and compare the stats of their players. As it suggests that disparity in opportunity is greatly reduced.This doesn't erase all the variables; nothing does. But you can make good assumptions by examining the stats in light of watching the players.
Frankly perusing statistical comparisons make you see better when you actually watch the players. You are less likely to overstate one great or terrible play, while understating 50 okay plays that have a greater overall effect.
I am an economist that manages a statistical (predictive) modelling and data analytics division.
Give me nearly any dataset, and what you would like it to indicate (within the context of the universe), and I can give you statistically significant/high confidence level findings (and dimension-based reports) that will demonstrate it... for a nominal fee, of course.
Statistics are a great tool, but they are (and should be) a single arrow in the quiver of assessment. I do agree that the analysis issue I raise *could* be partly due to the tactics employed, but I also touched upon the tactics that Pep prefers, which is to avoid the “last ditch tackle” scenarios altogether, hence why DMs that play within that sort of system often do not have as impressive statistics.
Pep has said many times that he welcomes performance analytics, and has encouraged the development of the “science”, but that it can be very difficult to accurately assess players within the context of quantitative measurement, for the reasons
@asahartford1 has alluded to. Pep has always championed new, groundbreaking methods of qualitative assessment, which integrates data points with educated minds and/or algorithms for analysing overall player tendencies, strengths, weaknesses, and more.
