New DeanO Pod

Home for all your discussion of basketball statistical analysis.
Crow
Posts: 6250
Joined: Thu Apr 14, 2011 11:10 pm

Re: New DeanO Pod

Post by Crow » Wed Sep 05, 2018 3:47 pm

Did you listen to the new link interview? Would you name at least one of the interesting predictions / comments?

schtevie
Posts: 362
Joined: Thu Apr 14, 2011 11:24 pm

Re: New DeanO Pod

Post by schtevie » Sat Sep 15, 2018 6:18 pm

Crow, apologies for the delay in reply. Not sure why I had clicked on the old link thinking it was the new one....Oh well. And, no, the new podcast contained no discernible predictions to my perception. But having listened agin to the old one, I can list the claims that were of note to me and (more or less) tested by last year's season.

(1) Beginning at 6:13, regarding the evolution of the three point shot... ""To some degree that market advantage has gone away...". I honestly have no idea what factual basis informs Dean's views on this point, perhaps it is very subtle. The fact of the matter (indeed one of the basic, robust stylized facts of NBA analytics) is that the returns of the three vs. the long two has been constant and enduring. And I mean really constant. Since distance data became available in 2000-01 the average 3P% has been 0.356 with a sdev of 0.015 and for the long 2 we see 0.399 and .0160, with difference expressed as additional points for a 3PA being 0.270 with a sdev of 0.010.

And then speaking to a kind of prediction, last year saw the second biggest increment in 3PA at the expense of mid and long 2s.

Perhaps Dean would explain his remarks on this topic, what came at the end of 50 (non-continuous) years since the introduction of the 3 point shot in professional (ABA/NBA) basketball. Switch four shots, get an extra point (basically). That this reality endures 50 years after folks should have been begun thinking seriously about things...there are no (positive) words.

(2) Beginning at 14:48 the topic of the Celtics arose, regarding their dramatic roster changes, IT, Kyrie, Jae Crowder and all that. Would they be an improved team?

Now, to be fair, predictions are hard and inherently uncertain and the 2018 Celtics off-season roster moves provide a very specific cautionary tale that folks should understand and remember that the standard errors on estimates of (most) player values aren't that small.

But it is worth revisiting Dean's remarks as I think the ex post consensus on the trade is largely incorrect.

Beginning at about 16:45, Dean makes and emphasizes the point that younger guys (rookies) are not going to be good. But then one turned out to be real good (not that the rookie component explains any net improvement of the Cs, but that is another story). Then there is the take that Kyrie for IT would be approximately even. I think that would have been right overall, abstracting from the terrible injury of Thomas. Kyrie performed exactly as expected and essentially no better on net than IT the year befor. Next, re Amir Johnson's departure (at about 20:24) it was remarked that there had been a decline in performance over years (though no mention was made of this having been an expected aging-curve related effect, as I recall). And then a curious characterization of the loss of Kelly Olynyk also a loss, one player who (by RPM) was good on the Cs and then got much better on new team.

Finally, at 27:26 there were the Jae Crowder remarks, where Dean O's comments had a strong dose of reversion to the Mean O, saying that he was definitely not a borderline all-star (as suggested by RPM-related hype) but rather the kind of player who is 3rd best on a championship-level team, or the 2nd best on a team in contention.

Two points to make about this. First, again, Jae Crowder's deterioration is a reminder that player performance is indeed variable, on rare occasions very much so. But second, I have no idea what Dean's accounting scheme is. If I look at the last five seasons (and I am willing to bet the story is the same forever) the 3rd best player on the championship team and/or the 2nd best player on the runner-up have an average RPM that is greater than the JC's offending 3.89 in 2016-17.

(3) Finally, to recall a conversation already having begun in this thread, at 21:56 we get the bit how coaches are worth 12 wins in a season. And I think it might be worth revisiting this in light of the Spurs' performance in 2017-18.

Here we saw Popovich unexpectedly lose his one superstar, facing a coaching condition he had not seen since his rookie year (when his starting mid-season job was just to tank so as to maximize the chance to get Tim Duncan). And how did "he" perform? Well...kind of as expected from my point of view,, with no evidence of 12 wins having been provided (though it isn't entirely clear how such evidence would be found). Per habit, he relied on his inherited roster (whose minutes from the previous season rose as a share from 65% to 86%) and their performance was essentially exactly as expected (in terms of RPM with aging effects). And then for the new players who played any substantive minutes (Rudy Gay and Joffrey Lauvergne) their performance (again in RPM terms) deteriorated, not outrageously so, but relative to aging curve considerations, so by this measure, there was no Pops magic in evidence.

We shall see how things go this year. In RPM terms (assuming the roster will be as currently listed on basketball-reference) the suggestion is that they will be a bit worse still. I calculate +2.2 for the team with no significant minutes adjustments (last year's minutes, coincidentally add up to 97% of a season, and I just prorated from there) but where now there are 37% of the minutes allocated to newcomers. Perhaps some Popovich magic will be realized, such as DeMar DeRozan becoming a new, middle-aged superstar. Let the games begin!

Crow
Posts: 6250
Joined: Thu Apr 14, 2011 11:10 pm

Re: New DeanO Pod

Post by Crow » Sun Sep 16, 2018 3:05 am

Delay not really a problem. If I get a response, that is a win, even in overtime.

I quickly listened to parts of new podcast but it didn't grab me that day.

I'd be interested in hearing reponses to your comments or anything Dean wanted to highlight from new podcast that I either missed or didn't light up on but perhaps could if nudged.

schtevie
Posts: 362
Joined: Thu Apr 14, 2011 11:24 pm

Re: New DeanO Pod

Post by schtevie » Sun May 12, 2019 9:44 pm

Following up now, with another Popovich coaching season in the books. Performing the same exercise of summing up player RPM changes for 2019 vs. 2018 and imputing all "unanticipated" changes to be the coaching contribution, what do we find? Well, about what any right-thinking person should expect...

As for the 2019 Spurs that continued on from 2018, their minutes increased from 53% to 63% of the total (a Popovich trademark). And their minute-weighted RPM increased by 2.17, collectively. Conversely, the newly-minted Spurs in 2019 saw their minute-weighted RPM decrease by -1.03 (again a familiar story). This then is a net change of +1.15 that represents the "coaching contribution" (and given the particulars, it doesn't look like net aging effects would change this estimate much at all.)

And what does +1.15 imply in terms of wins? Well, owing to the recognized "players don't play as hard from in front" RPM effect, the actual effect on the scoreboard would be a fraction of this. The 2019 Spurs roster NRtg was +1.7 vs. a roster RPM of +2.09, what is a ratio of 0.81. Applying this we might infer that Popovich was responsible for a NRtg of +0.93.

Conveniently then, tucked 0.9 points below the Spurs in the 2019 NBA NRtg standings is the Orlando Magic, whose predicted wins were 43 vs. the Spurs' 45. So, unless there is some egregious error above, this method comes up with +2 wins being the Popovich effect.

Finally, I feel obliged to reiterate (again) that I am not anti-Popovich. To the contrary. He is interested in quality wine, gets impatient with dumb media questions, tolerates little bullshit, and is aspirationally woke. All good things...and he's a good coach! Just not 12 wins good.

Crow
Posts: 6250
Joined: Thu Apr 14, 2011 11:10 pm

Re: New DeanO Pod

Post by Crow » Mon May 13, 2019 12:22 am

That might be his net new 2018-19 marginal impact; but to be fair, some of his coaching impact is already baked into players by the previous season.

Post Reply