What Does This Team Need Most?

[TheMobb]

Actually, I’m going to say the Warriors keyed on sagging the defense to the paint and not necessarily on Domas first. And why wouldn’t a team do that with Domas stuck in the paint to be effective at the time, Fox who was a paint monster and the Kings up and down 3 point shooting.

I was commenting on what the Warriors assistant coach said, that their priority was stopping Domas as a hub and distributor and blowing up the DHOs. Watching the games you could see that and also that they were conceding the 3 to Fox. I’m sure that had an effect in bending the defense and opening things up for the Kings guards.

To compare it to the Warriors-Rockets series, the numbers paint that Jalen Green had a terrible series. However you could see the Warriors constantly blitzing him and that he was the priority to stop, that leaves the others on a 4 on 3 but it’s up to the others to make the right reads and knock down shots.

 

[Insomniacal Fan]

I assure you that I am not purposefully ignoring it. My position, which I have stated, is that we are looking at a very small sample size (7 playoff games since he has been "Domas") and we should not believe that those 7 games have remotely the same predictive power as the 246 regular season games he has played as a King.

I recognize that he is not a good defensive player (but a GREAT rebounder, which is usually considered a component of defense statistically), but he remains wildly successful in advanced stats that include both offense and defense. He has a huge +++ impact on winning, even if he is a weak paint defender.

And, to address the remaining point, I believe that "easy to scheme against" is just a narrative to explain the difference between his large sample size stats and his small sample size stats. The idea that NBA coaches don't actually do any coaching until the playoffs start is not credible to me. And, of course, if he were statistically better in the playoffs than the regular season, the other side of the argument would call him "uniquely built for playoff basketball". I would endorse neither of those views.

The small sample sizes criticism is interesting to me. There are statistical tests that compensate for differing sample sizes between groups, so it should be possible to determine if the playoff effect is real or not

I plugged Domas' game logs into a one way ANOVA analysis. Comparing 246 regular season games as a King, to 7 in the playoffs, there were notable p-values for TS%, DRB%, and AST (There were also a few other stats that are derivative of those)

ANOVA [TS%] results: F=7.3666, p=0.0071
ANOVA [DRB%] results: F=9.8929, p=0.0019
ANOVA [AST] results: F=4.4996, p=0.0349

I think we can say that despite the small sample size, there was a statistically significant effect between Sabonis' regular season efficiency, and his playoff efficiency. I'm not a statistician, and I would appreciate any criticism of this approach; but I believe this debunks the small sample size objection to the extent I could publish these results in a scientific journal.

Now, looking at Sabonis' broken hand stats, I don't see evidence that he was hampered by his avulsion fracture; in fact his best statistical run came in the months after the break. (As an aside, this jives with my personal experience. I had an avulsion fracture on my toe once, and aside from occasionally feeling sudden-random-numbing pain, I was pretty functional after a couple of weeks.) Domas compensated effectively.

I have considered the idea GSW was a bad matchup. The stats between the regular season games against GSW and the playoffs still show a significant effect on DRB%, but there isn't enough evidence to prove that Domas' TS% was affected by playoff basketball relative to an unfavorable matchup vs GSW.

I haven't disproven the stomp hypothesis (yet).

 

[Turgenev]

The small sample sizes criticism is interesting to me. There are statistical tests that compensate for differing sample sizes between groups, so it should be possible to determine if the playoff effect is real or not

I plugged Domas' game logs into a one way ANOVA analysis. Comparing 246 regular season games as a King, to 7 in the playoffs, there were notable p-values for TS%, DRB%, and AST (There were also a few other stats that are derivative of those)

ANOVA [TS%] results: F=7.3666, p=0.0071
ANOVA [DRB%] results: F=9.8929, p=0.0019
ANOVA [AST] results: F=4.4996, p=0.0349

I think we can say that despite the small sample size, there was a statistically significant effect between Sabonis' regular season efficiency, and his playoff efficiency. I'm not a statistician, and I would appreciate any criticism of this approach; but I believe this debunks the small sample size objection to the extent I could publish these results in a scientific journal.

Now, looking at Sabonis' broken hand stats, I don't see evidence that he was hampered by his avulsion fracture; in fact his best statistical run came in the months after the break. (As an aside, this jives with my personal experience. I had an avulsion fracture on my toe once, and aside from occasionally feeling sudden-random-numbing pain, I was pretty functional after a couple of weeks.) Domas compensated effectively.

I have considered the idea GSW was a bad matchup. The stats between the regular season games against GSW and the playoffs still show a significant effect on DRB%, but there isn't enough evidence to prove that Domas' TS% was affected by playoff basketball relative to an unfavorable matchup vs GSW.

I haven't disproven the stomp hypothesis (yet).

I do not claim to be a statistician either. But, trying to be fun:

It would be better to post your question first, define your outcomes of interest, then report the effect estimates for each outcome. Rather than performing a test on several outcomes and then reporting only the P values for outcomes that were significant and saying you have something ready to submit for peer review.

One way ANOVA is for two independent groups. Since Domas produced all the stats you have analysed, they are not independent samples. So this isn't an appropriate analysis method to use.

P values are for chumps.

When choosing which analysis method to use, you do have to consider other variables that can influence the outcome beyond just 'regular season' or 'playoffs'. These can include opponent and injury. Maybe a linear regression? But how easy is it to define 'injured' or 'stomped on' for an analysis like this?

I feel the best way is just to eyeball the stats and to acknowledge the limitations. That is, Sabonis' output seems to be less in the playoffs, but it is a small sample, he has played hurt, and Looney was pushing off all series. And maybe he wouldn't have been stomped on if he was terrible.

Then maybe say hey Sabonis might or might not be that great in the playoffs, but we are Sacramento and we don't get to go to the playoffs that often, if at all. What we really need is a good regular season team to get us to the playoffs more often.