Tatum's on/off differential per 100 possessions is +28.1. That is absolutely astronomically, especially 12 games in. For comparison, Lebron at his absolute peak (08/09 in Cleveland), was at +21.2. Theis has the next best on/off per 100 on the team at a much lower +13.4. Tatum and Theis are also around +16 (per 100 possessions) on the season, with Brown and Walker nearly 8 points lower each at +8.7.
Tatum's value is insane where it matters most. And it is that way because there are no credible back-up PF's on the team. Thus, he might not be the best player on the team (that is still Walker), but Tatum is the most important player on the team. We lose Tatum for an extended period of time and the team is in real trouble.
Again that's a testament to his situational value on the Celtics, not how good he is as a basketball player. On/Off doesn't account for the strength of the units he's mainly playing in, the strength of the opponents he's playing against, the nature of the game when he's on the court (blowout, tight game, etc.) and many other factors. Impact metrics that adjust for such factors such as RPM and APM are much better when it comes to gauging a player's value, and Tatum hasn't been setting the world on fire in this family of +/- stats from what I've seen (I took a look at his RPM on ESPN last season and they weren't anything special).
You know what else wasn't special last year, his +- and on/off numbers (though like this year he had some of the higher ones on the team).
It is not a coincidence that the best 5-man units for the Celtics this year, all have Tatum in the lineup. That holds for 4-man, 3-man, and 2-man as well. Tatum has been the one constant in Boston's most effective units all season long. He is Boston's most valuable player in the only area that truly matters.
It MIGHT not be a coincidence, but it is certainly possible it is. The sample size on this plus minus is too small, to dependent on who he's playing with, too dependent on who he's playing, too dependent on too many other factors for us to draw any certainty from the numbers yet. I've seen to many games this season where he ends up +10 to 16 but actually played pretty bad for me to put too much stock in his plus minus.
Now all that being said I also dont think he's been as bad as his efficiency numbers would indicate. Some of that is just him missing shots that I think he will indeed make, shots that are within his current ability to make. So basically he's better than his peripheries indicate, but not s good as the plus minus indicates which for a 21.5 year old is good.
best isn't just +-, Tatum has the best net rating on the team (16.5) and is in all of the best Net Rating units as well. Pretty much any metric you use thus far this year and Tatum is in the best lineups of that metric. For example, of the 2 man groups with at least 100 minutes, Tatum/Hayward has the best net rating at 22.3. Tatum/Theis is 20.3. Tatum/Brown is 19. Brown/Theis is 4th at 17.8 just ahead of the Walker/Theis unit of 17.7. Then you get Tatum/Walker at 16.5. Then Theis/Hayward and Tatum/Smart.
Tatum is thus in the top 3 2-man groups for net rating. In fact, the first time you don't have Tatum or Theis in a 2-man group is the pairing of Hayward/Walker which is the 9th best pairing. The same is true of 3-man groups. There are 11 3-man groups that have played at least 100 minutes together. Tatum is the only player in each of the first 4 groups by Net Rating (the top 2 also have Theis). Smart, who seems to be all the rage on this board of late, doesn't show up until the 6th best 3-man group (with Brown and Tatum). The trend is similar with 4-man and 5-man groups.
So in sum, no matter how you look at it, Tatum is the player that is in all of Boston's best and most productive units this year, which in my mind makes him the most valuable player on the team. By the way, this a similar argument you could have made with Embiid in the Toronto series last year i.e. you look at his stats and he underperformed, yet the Sixers were something like +110 with him on the floor and -125 when he was on the bench. Sometimes players just have a value that isn't reflective in the raw stats. I'd argue that Tatum, at least thus far this year, falls into that category.