Is that stat really 'noise proof' though? Look at this post in the Lindelof thread:
https://www.redcafe.net/threads/victor-lindelöf-2022-23-performances.471408/page-29#post-30209897
From those stats, you would think he should be comfortably one of our first choice centre backs. I'm assuming the stat you're referring to is something to do with winning the ball in the air or rather getting the first touch right? However, does that take into account what the action of the opponent winning that header actually created? What I getting at is that from actually watching the game, Martinez is twice the player Lindelof is aerially and it's due to intangibles or rather stats that simply do not exist e.g aerials contested that led to an advantageous action to the opponent, aerials contested by the defender in the least advantageous situation, aerials contested that won 'momentum', aerials contested that was of significant importance due to the chance (like the xpgzy or whatever % that we see nowadays) etc etc. There's so many variables that can go further into detail if we really want to examine what 'winning' aerially means.
Now I'm not trying to defend Martinez and excuse statistics because I agree circumstances such as playing in a team with teammates that suit his playing style obviously massively help (as with any other player) but I don't like stats being used to definitively say 'xyz is not good in the air'. It should be more accurate to say 'he's specifically poor in or doesn't win the first aerial ball' as those are two completely separate statements, which can shape a discussion or narrative. Another example is saying one player is a 'good' passer because of their high completion percentage but it means nothing compared to another midfielder, who passes forward and from much more difficult positions. The difference is that we have stats like 'progressive passes', 'distance moved', 'passes in final third' etc, which can further help contextualise a point; they're not perfect representations to judge a player or make a statement but at least there's some variation. With aerials, I don't think there's as many stat instances to give different information.
TLDR: Effectiveness and 'winning' a stat, although related can be two very different things.