Doctor MJ wrote:OhayoKD wrote:on/off tends to inflate the actual effect of 'fit" issues. Bad fit and all, Luka has basically averaged jordan/magic/hakeem game-level lift over a bunch of >10 game season samples for his first 5 years, and we know his ratings are basically unaffected with bench players and without starters.
Clearly you're talking WOWY-type stats yes? Those are worth looking at certainly, but using them to dismiss more granular +/- stat is not something I'd do lightly.
If you want to elaborate on this, that might help give me more confidence.
Well you saw the tweets posted where Luka has the best net-rating with bench and the best net-rating without 3 starters. He also plays alot less with starters, so you have a potential explanation for the game-level vs on/off gap there, not to mention him playing with other capable ball-handlers for most of his career.
And then there is the matter of this phenonom taking place with a very similar(offensive) player in Lebron in Miami where his on/off plunged staggering with another elite playmaker but his WOWY was still great(though admittedly not the girnormous outlier he was with better fitting cleveland rosters as a small-forward) and then in the playoffs they converted two titles(facing 2 atg level opponents) despite a bunch of injury context.
If this is what it is for Luka, then his on/off doesn't really matter to me. If the power of rotations isn't preventing you from taking otherwise bad teams to contention(in this case i think playing okc or minny close would qualify), then the team not being bad whenever you miss a few minutes of game time doesn't really affect how I view you as someone who elevates teams or the variety of players you can mantain that elevation with. The Mavs, even with luka playing with other playmakers got alot better with Luka over games so if the on/off is just a result of staggering and that staggering does not inhibit you from making the team much better...does it really matter?
Mind you this isn't specific to a certain playstyle. Curry has a good backup guard...suddenly negative on/off. Duncan and Drob, Jokic in the playoffs when he staggers with Gordon. The first thing i think you should look at when there's a disparity between the game and the spot minutes is what the rotations are like imo.
I'm also just very confident in Luka's skills and his film based on the impact we've seen from similar players and how he takes out most of a team's defenders on his own at a frequency no one else in the league does I think. I would need alot more focused and film-backed argumentation to buy the on/off is spotting something the game-level stuff isn't.
OhayoKD wrote:The NHL used this stat for a long time..
NHL's +/- is a completely different stat from basketball +/-. NHL +/- is just tallying up chosen positive and negative actions and subtracting the latter from the former. It's basically a less discriminatory PER, not an impact stat.
There's truth in this statement traditionally, but once basketball people made RAPM-type stats, the hockey people followed suit, so that data certainly exists now.
Having done some hockey +/- analysis, I'll note a couple things:
1. The low scoring of hockey makes +/- a noisier stat, and this is something that got a lot worse in the modern game when they let goalies where enough pads to basically cover the entire net.
2. Note that this also makes hockey a noisier game and one more like baseball where the playoffs don't even necessarily say that much about who the best team would be if you held all the same series again. With baseball, the main focus of analytics has been from a production/player tracking perspective rather than impact metrics for reasons not unrelated to this. Hockey is a challenge because it has similar noise issues to baseball, but as a flowing field sport it's a lot harder to measure production/player tracking beyond the absolute basics.
3. One wrinkle I didn't really grapple with until I started looking at the data was the fact that defensemen just play significantly more than the center and wings. I knew that +/- was a stat associated with defensemen and figured it was just because goals and assists tended to be dominated by the more offensive positions, but it's also because of them playing greater minutes, and thus having a tendency to have the bigger raw +/- numbers.[/quote]
Did not realise there were two plus-minuses but cool that they're gravitating towards winning as well. A big difference between soccer and hockey is the 82-game regular season sample. Do starters miss enough games to draw inferencers from there like we do from basketball?