For me, I personally think a lot of advanced metrics fail to hit the mark for me, but I totally reject win shares, even though I think it’s a cool idea in theory. Fundamentally, my problem is that it fails to account for the team structure around a measured player as well as the strength of teams they play, and, as such, it results in some pretty strange outcomes.
For instance, the leader in playoff win shares in 2020 was Anthony Davis over Lebron James. Lebron had splits of 27.6/10.8/8.8 with 1.2 steals and 0.9 blocks. AD had splits of 27.7/9.7/3.5 with 1.2 steals and 1.4 blocks. Their stats are almost identical, but Lebron had significantly more assists, yet he has a slightly lower total win shares and WS/48. I feel like this is weird considering the general consensus after the bubble was Lebron was the clear cut best player that postseason, especially after winning Finals MVP.
A more egregious example in my mind is 2011. That year, Lebron led the league in total playoff win shares, even though he ended up losing the Finals to Dirk Nowitzki, who, himself, may have had one of the single greatest postseason runs ever, beating Kobe and the defending champion Lakers, the Duncan trio Spurs, and the Heatles all in a singular postseason, all without being on a team without a ton of talent. Wouldn’t being one of the primary stars on a championship team bias the stat towards him getting more win shares? Shouldn’t that be the case given what the stat intends to observe? And how can Lebron have such a high amount when he’s sharing the court with 2 other great players?
What about you guys? Are there advanced metrics that you think miss the mark in their attempts to further an understanding of the game?
Win Shares are the best we can do for players before 1973-74. But there are better all-in-one stats, for sure. Win Shares is especially suspect when measuring defense, but then most stats are suspect when measuring defense, even the advanced stats.
PER has absolutely no reason to exist except that it came before many other advanced stats and was well publicized by ESPN. But there are many better stats that cover the same time period. PER gives too much credit to inefficient volume shooters. If someone cites PER as their advanced stat of choice, I know they aren’t to be taken seriously.
All of the all-in-one stats have flaws. None of them should be taken as gospel without examining what it does and does not measure. But if it produces surprising results, the key is to examine why the results are surprising. Maybe there’s a player we’ve overlooked – or maybe it’s a flaw with the stat.
Even True Shooting Percentage, which is pretty trustworthy, assumes that 0.44 is an estimation of how many possessions one free throw is worth, which is an approximation. Furthermore, a great True Shooting Percentage on wide open uncontested catch-and-shoot shots is different from the same or maybe even a slightly worse True Shooting Percentage off the dribble on heavily contested shots.
So context still matters. Someone who plays with Jokic and benefits from his passes may not be nearly as efficient on a lesser team.
I agree with your point on TS. At this point in the development of the nba, how do we not have an actual points per possession metric? A 3 and a 2 + 1 Ft are the same. A 2 and a 2 + 1 missed FT and 2 made FTs are the same. Etc. But TS rates those differently. It should be so easy to record and tabluate but instead we have to guess.
i mean they do have it, it’s just under pay-walled tracking data