Bob8 wrote:Doctor MJ wrote:Bob8 wrote:
You're using only raw +/- in this whole conversation. On/off is as raw as it gets stat too, enormously influenced by who replacement player is. You brought, talked and made big conclusions using only raw +/-. You could have used regression, adjustments or whatever you wanted, but you didn't. We're focusing on stats you have brought in conversation not me.
If you measure players impact and say player A is far more impactful than player B and all that changes in a moment, because playoffs environment is more basketball pure. Than I can only say, you should focus more on other aspects of player's game, when measuring true impact of a player, because your conclusion based on +/- is obviously wrong. Nobody can become good or bad overnight. The main difference in playoffs is that all competitive teams are using their best lineups as much as possible. You can have a team with big problems with injuries in RS suddenly playing with full rosters, of course their performance would be totally different. Players at this point of the season are what they're, but lineups in playoffs can be very different than average lineups team has played in RS. Just looked at 3 games OKC and Mavs played. OKC with full roster all 3 times, Mavs missing half of the roster in game 1 and without Luka in game 3. The only time they played with full roster, it was blowout win for Mavs. Normally that extremely healthy team as OKC will have more wins in RS than extremely unhealthy team as Mavs. But in playoffs that extreme RS health won't matter much. And I don't believe many will be surprised if Mavs win potential series against OKC.
And that's exactly what I'm saying. All Booker's greatest, or should we say it impact, disappeared, when CP3 couldn't assist his game anymore. Suns lineup failed and with that Booker too. Booker was still the same player, but his team wasn't playing the same anymore and his +/- has fallen off the cliff. If he was truly more impactful player than Luka, that shouldn't happened, he should lead his team no matter how CP3 plays. Booker is just another prime example how meaningless +/- is. His +/- impact is far superior to Luka's in last 5 years, but if there was a trade between Suns and Mavs, Suns would need to include numerous FRPs to even start talking. It seems to me that not only Luka's stans consider him as top player in Nba, even though +/- is not agreeing with it.
I'm referencing raw +/- and On/Off yes, but I've made clear repeatedly why I'm doing this and you keep attacking it like I haven't countered what you say.
If what you're wanting is to compare RAPM data, I'm generally happy to go there. Certainly if that data presents a different story than what I've mentioned it should be brought up.
Why do I generally reference more raw data?
1. More understandable to a broader audience.
2. Standardized and readily available for free on nba.com & b-r.com, which RAPM isn't.
I do understand the challenges with the approach I use in that it makes sense for someone to counter raw +/- with on/off concerns, and counter On/Off data with concerns that speak to the need to use things like regression. But once I've addressed those issues, it doesn't make sense to keep bringing them up like I haven't.
So either:
1. You're understanding that I've addressed your concerns and you're purposefully pretending I didn't.
or
2. You're not understanding something about what I've said previously, and it's not triggering recognition on your part that you really need to re-consider the things you think you understand that you don't.
My inclination is to say (2) because I would prefer to assume good faith in those I debate with.
If it is (2), then I would say you should be looking to ask more questions, and do so with sincere curiosity.
Re: nobody becomes good or bad overnight. Right, but they can become impactful over night, and impact is what drives value, which is what the Most Valuable Player Award is named for.
You keep talking like the fact that +/- is dependent on context is a bug, but when talking about value-based achievement, I consider it a feature.
What about 3rd possibility?
- You didn't address concerns good enough.
I'm going out so I don't have time to go over everything again. I didn't get good answers,
How/why SGA became +/- monster overnight, why Curry has fallen of the cliff, how/why KP became a +/- monster when joining great team, why Tatum has very modest on/off. Why Luka is after trades crucial for Mavs D...
My quick explanation for all those questions is that teammates are crucial for having elite +/-. That makes very problematic comparing impact of players that are playing for different teams, especially between teams with great rosters and teams with very average ones.
Concept is robust only if works and brings good results under a variety of conditions. That doesn't look to be the case for +/-.
If I didn't address the concerns good enough, you should still be making clear that I tried to address then should be looking to focus the conversation to try to get clarity. Instead you're literally just saying the same stuff over and over again like it's the first time.
To some of your specific things - which I've certainly talked about some, but I've also emphasized that this isn't a situation where if I can't explain the numbers it makes sense to ignore them:
First you're focused on the idea that rapid change in stats is a flaw because the player in question is still the same guy, and as I've said I see impact as a thing that can change rapidly which you disagree with. That point of disagreement is probably the most fruitful thing we could look at to understand each other better.
Okay: How Shai +/- monster overnight. I'd point to the fact that he's shooting better from everywhere for offense, and he's making more plays defensively. In both cases the presence of teammates are probably helping him, just as his presence is helping them. This is the kind of synergy you're seeking when you are trying to optimize fit, and everyone involved is fortunate for it. I don't personally believe that fit should be something we attempt to normalize for in an MVP setting, but reasonable minds can differ.
Curry fallen off the cliff. This is one where I look forward to see '23-24 RAPM studies that have large sample because it may not actually look like falling off the cliff.
In a nutshell, in many years Steph plays more with Dray than anyone else, and those are two great flavors that go great together. This year his two main minutes partners have been Klay Thompson and Andrew Wiggins. Now, a couple years back that doesn't sound so bad maybe, but both have been universally acknowledged to be awful this year compared to their past standards, and if we break down the On/Off's of the 3 guys:
Steph's Offensive On/Off: +3.2
Klay's Offensive On/Off: -1.1
Wiggins' Offensive On/Off: -0.2
Steph's Defensive On/Off: +5.3
Wiggins' Defensive On/Off: +6.7
Klay's Defensive On/Off: +8.0
Green & red coloring used here to emphasize what's good and what's bad. You can see Curry's looking quite good by this compared to his main running mates. This leads me to generally conclude:
We should hold off before we assume that more sophisticated metrics will see this as a case of Curry having much less impact than before.
Not that I'm saying there aren't other indicators - his box score is down basically across the board and his age is up to 35 - but this isn't a case where it seems like he's gone from elevating his teammates' raw +/- to stifling them. If a rising tide lifts all boats, that also tells us something about the fall.
Re: KP became a +/- monster on a great team. Well so clearly what you mean by this is that his raw +/- went up when he joined a team full of guys with high raw +/-, and there my answer would be the same as yours.
I don't see KP as a "+/- monster" though. If we just go by the On numbers (which are per 100 possessions), here's how the main Boston Celtics by descending MP look this year:
Tatum +12.1
White +13.6
Brown +9.6
Holiday +10.2
Horford +11.7
Pritchard +11.1
Hauser +12.8
Porzingis +11.7
So, this just generally looks like Porzingis is working well in Boston, but he's not showing indicators of irreplaceability that his teammates don't have. All these guys do deserve props for functioning so well in lineups out there, and basically as a rule that means that their value is higher than what we'd expect them to have in most contexts.
I'm happy to get into further discussions on those guys numbers if you want, but I'll say that the fact that they're so close makes it harder to infer causes.
Why doesn't Tatum have a bigger On/Off? Well, the first level of truth is simply that the team has done well enough without him that despite the Celtics doing incredible with him, On/Off isn't that big. Then of course there are the questions of how the heck the Celtics being able to do have such depth of dominance, which is no simple thing to answer. You can't simply keep adding "good players" to a basketball team and expect them to become dominant like this. A lot of great GMing & coaching going on on that team
For the record: I'd say I'm not a Tatum-skeptic. I generally rate him higher than most in competitive achievement. I'd say thought that if we were measuring guys by how compelling I find them, he'd rank lower and more in line with most folks.
Re: Luka after trades crucial on D. I mean, it's small sample and we're in the dog days of the season where everyone's just waiting for the playoffs. I'm not drawing any big conclusions yet, let alone something about Luka's defense.
To your last of +/- problematic for judging impact, I think this gets back to the whole thing where I see impact as a much more fluid thing than you do, and I think frankly it's because you don't really separate it out in your thinking from goodness. If would be great to be able to accurately talk about "impact players" who have a universal additive effect of +X.0 impact in all settings, but we just can't. Game's too complex for that.
"
