RCM88x wrote:lessthanjake wrote:RCM88x wrote:
Okay so initially it appears you're comparing two completely different numbers and trying to pass one off as something it's not to make the argument more convincing.
If you're explaining correctly, those ratings do not incorporate Houston's expected team strength, making them effectively worthless. You can't just give opponents their net rating from the regular season and not also give Houston their own, right?!
For example. If a +10 team plays a +1 team to a (+1) point differential, that is not be a +2 adjnet rating, that should be a -8 adjnet rating.
Let apply this to 2013:
Houston (+3.6) plays OKC (+9.8), to a net of (-6.2)
Adjusted net should be -3.6+9.8-(-6.2) = 0.0
It should NOT be 9.8 - 6.2 = 3.6 (or 3.7 because of rounding I'm assuming).
You cannot just ignore expected team strength when calculating relative adjusting net ratings, that is just silly.
If we're going to compare a regular season points differential to this ""relative"" point differential we need to do that for every team now and see what the results are. I am going to feel pretty confident that Houston's performance is still below average here. We should do the same comparison with the Warriors, Cavs, Spurs, Raptors, Celtics, etc... from the same era and see what those results are (I do not subscribe to this database so I cannot).
EDIT: Just to display more of how silly this is.
'16 Warriors (+10.7) played '16 Cavs (+6.4) to a (-0.6) differential
According to this Thinking Basketball stat as you explained it, Warriors are apparently a +5.8 adjusted rating (6.4 - 0.6)??
Absolutely not, that should be a -3.7 (-10.7 + 6.4 -(-0.6)) adjusted rating because they significantly underperformed what was expected based on their existing team nets.
You’re just shifting numbers around. The playoff relative net rating value I gave is aimed at telling us how well the team did, while adjusting for the strength of the opponent (which is a pretty standard way of assessing how well a team played in the playoffs—hence why Thinking Basketball provides it). I am then reporting out that number alongside the Rockets’ regular season net rating, so you can compare them and see which one is better. All you’re proposing is to take the two numbers I provided and subtracting one from the other, so that we can specifically get a measure of how well the Rockets did in the playoffs (adjusted for strength of opponent) compared to their regular season performance. But you can literally just get that number by taking the playoff number I provided and subtracting from it the regular season value I provided. And the discussion I gave of the numbers explicitly compared the playoff and regular-season values, so I was effectively making that same comparison.
Anyways, if you really want that number, then the numbers would be the following (with a positive number meaning the Rockets’ playoff relative net rating was higher than the Rockets’ regular season net rating):
2012-13: +0.1
2013-14: -0.2
2014-15: +0.8
2015-16: -8.6
2016-17: 0.0
2017-18: -2.3
2018-19: +3.7
2019-20: +1.5
So yeah, again, if we look at net rating, the Harden Rockets did better than the regular season in 4 out of 8 playoffs, and did exactly the same in one playoffs. In the other three years, they (1) barely did worse than the regular season (-0.2 in 2013-14), (2) had a negative value in 2017-18 that was a positive number until CP3 got injured; and (3) had one year getting destroyed by the Warriors in a series that the Rockets had +4000 betting odds in.
I don’t think it’s a fair portrayal to look at those numbers and say the Rockets underperformed in the playoffs. And the same is true if we look at how they did compared to what the betting odds were in all the series they played. They won every series they were significant favorites in, lost every series they were significant underdogs in, and went 2-2 in the four series where the odds were very close. Just like the net rating data, this looks like a team that did pretty much exactly as well in the playoffs as we’d expect.
Shifting numbers around? I'm explaining to you how SRS works and why this adjusted number (which is the definition of shifting numbers around) is silly. I don't care what Thinking Basketball provides if it's a silly number relative for this comparison.
Okay, so now we have proper adjusted net ratings. Overall, really only one season they performed meaningfully better than expected (2019). and underperformed in the prior 3 years (love how somehow we're using 2013-2015 numbers now too for some reason despite those being completely different team construction, coaches and league environment, and clearly not within the scope of this discussion but whatever.
Then... betting odds?? How is that relevant at all? Those odds are not set based exclusively on regular season performance indicators, they have expected playoff underperformance and overperformance baked in. That's why the 2018 Rockets weren't favorites against GSW in 2018 despite having superior RS indicators. Besides, being an even 500 (if we use your categorizations) against betting odds kind of proves my point no, a good playoff team probably would be expected to overperform betting odds (unless they are favorites in every series natrually), not perform as expected in them.
Okay, as an initial matter, I’m not sure why you’re acting like I don’t know how SRS works, and I’m not even sure why you’re talking about SRS specifically. I certainly don’t need an explanation of SRS, and it’s worth noting that this conversation actually is not even about SRS, which is a subtly different concept than net rating. SRS is average MOV per game adjusted for opponent strength, while relative net rating is net rating per 100 possessions adjusted for opponent strength. These are very similar concepts but not quite the same, since there’s not 100 possessions per game.
Nor am I sure why you’re acting like the data I provided wasn’t directly relevant to the question at hand. It absolutely was, since all you had to do to get the number you wanted was to subtract the values I specifically put right next to each other for each year. Like, I made a post specifically geared towards reporting out regular-season and playoff data in an easy-to-compare way, and you’re objecting that I gave you “silly number[s]” because…you wanted the two numbers I gave for each season to be subtracted instead of simply put side-by-side??? Like, surely anyone reading my post is able to look at something like “2012-2013: +3.6 RS vs. +3.7 playoffs” and figure out that the Rockets’ relative net rating in the playoffs than year was 0.1 higher than their regular-season net rating, right?
In any event, if you want to talk specifically about years with relatively similar team construction, then we should be zeroing in on 2017-2020 (i.e. after Dwight Howard), during which the average is clearly positive (i.e. they did better overall in the playoffs in those years than in the regular season). But I provided all the data from Harden’s entire time in Houston, in order to be as complete as possible.
Also, saying they “underperformed in the prior 3 years” before 2019 is not true since they neither overperformed nor underperformed in 2017. Anyways, I really don’t think anyone could look at this data and conclude that it suggests the Harden Rockets were meaningfully worse in the playoffs than in the regular season. They were pretty much exactly as good in the playoffs as the regular season. Granted, the teams that win titles almost always do *even better* in the playoffs than in the regular season, but the Rockets did basically tread water. They were not at all like the DeRozan Raptors or something.
