Tatum +12 in the 1 point loss. At some point this stops being a fluke and starts being the reality when it comes to Tatum and his actual value to the scoreboard. I mean he can shoot 1 for 18 and be +16. He is amazingly valuable because of who actually comes in for him.
Not sure if you caught the local broadcast but Sean Grande and Scal were making fun of this stat, basically called it garbage with proof being Tatum's great +/- while going 1-18 and being a +15, at the time in this game while having 4 points and not playing well.
To be fair he can produce crazy numbers in the +/- family of metrics (the ones that adjust for teammate quality, opponent strength, etc, not these on/off splits that take the scoreboard at face value) if he played defense and passed like the second coming of Kevin Garnett, even if his scoring was anemic. The thing is that his complementary attributes are nowhere near that level, his best complementary skill in help defense is just solidly above average.
I don't hate the stat, just stating that Grande and Scal think it garbage. One said, and I paraphrase, it's too early to try to gauge a player's play by that stat, then the other said, heck, after a season it's too early to use that stat. Neither seemed in love with it.
It's a noisy stat and can be deceptive. Tatum leading the league in it is extremely deceptive but it's just an 12 game sample. It should even out some by mid-season.
I wasn't mentioning on/off, I was talking about its more accurate cousins in RPM and APM. Those stats account for the factors I mentioned above, and they don't show anything special about Tatum that doesn't show up in the box score imo in his first couple of seasons (they're not available for this season since they require a pretty big sample size). But yeah plus minus stats need a big sample size to show any correlation/trends, and even then the correlation might be pretty weak/noisy.
APM is okay, though I think it a stat better used over multiple years rather than parts of a year or only one year.
I have no respect for the RPM stat. Anyone that won't publish a formula for a stat is hiding something and given the wanky results RPM can put out, I don't like it.
Almost all impact metrics put out wonky results every now and then, it's how you interpret it that matters. For example I don't think you'd rate Danny Green as a better player than Klay Thompson last season even though his PIPM was better, you'd just rate Green as an amazing role player who provides massive value on stacked teams due to his additive shooting and defense, while Klay's value is muted in such a role because he's not a top of the pack defender, and he doesn't provide a ton of complementary skills other than his shooting and non elite man defense, but he'd flourish if he had a larger offensive load and less of a defensive burden so he can focus on man defense and utilise his offensive arsenal more (he's actually pretty good at creating off the bounce and making a move off the catch, as opposed to Green's purely catch and shoot game).
And APM is better over multiple years, but even one year data is better than raw on/off. Anyways my point was that Tatum is nothing special in more advanced plus minus statistics that account for a variety of factors that contribute to the noise in raw on/off.