APBRmetrics

Posted: **Wed Dec 14, 2016 4:18 pm**

Biggest drops in expected wins since initial release:

LAL: -16.16
ATL: -13.86
BRK: -12.68
DEN: -7.65
DAL: -6.27

Biggest increases:

GSW: 18.09
TOR: 14.75
HOU: 10.56
CLE: 9.45
MEM: 7.59

16 teams are within +/- 5 expected wins compared to initial release.

Posted: **Thu Dec 22, 2016 5:36 am**

Through 12/20 games

Teams currently in playoffs according to standings who aren't in top 16 by these ratings: NYK, IND, POR

Teams not currently in playoffs according to standings who are in top 16 by these ratings: WAS, MIL, DET

At the moment, POR, SAC and DEN tied for the right to be annihilated by GSW in the first round.

Unweighted:

Code: Select all

Team  Rating Exp Wins
GSW	10.28  66.22
TOR	7.11	59.84
LAC	6.34	58.04
HOU	5.76	56.66
CLE	5.73	56.57
SAS	5.27	55.43
UTA	3.45	50.69
BOS	1.86	46.30
CHI	1.23	44.52
DET	1.01	43.89
CHA	0.92	43.63
MIL	0.68	42.96
OKC	0.20	41.57
MEM	-0.75	38.87
ATL	-1.29	37.33
WAS	-1.43	36.92
SAC	-1.53	36.63
DEN	-1.54	36.61
MIN	-1.82	35.82
IND	-1.90	35.60
NYK	-1.93	35.51
POR	-1.94	35.48
MIA	-2.22	34.70
NOP	-3.07	32.34
DAL	-3.77	30.46
LAL	-4.09	29.59
ORL	-4.24	29.21
PHX	-5.24	26.63
BRK	-6.08	24.58
PHI	-7.01	22.41

Weighted:

Code: Select all

Team  Rating Exp Wins
GSW	10.29  66.24
TOR	7.10	59.80
LAC	5.90	57.00
HOU	5.82	56.79
CLE	5.59	56.24
SAS	5.10	55.00
UTA	3.36	50.44
BOS	1.96	46.56
CHI	0.96	43.75
MIL	0.84	43.41
DET	0.80	43.28
CHA	0.78	43.24
OKC	0.07	41.21
MEM	-0.65	39.14
WAS	-1.25	37.44
SAC	-1.48	36.79
DEN	-1.54	36.60
ATL	-1.57	36.54
IND	-1.68	36.20
NYK	-1.83	35.78
MIN	-1.86	35.70
POR	-1.92	35.54
MIA	-2.31	34.45
NOP	-2.96	32.65
DAL	-3.61	30.88
ORL	-3.96	29.94
LAL	-4.26	29.15
PHX	-5.12	26.93
BRK	-5.91	24.98
PHI	-6.64	23.24

Posted: **Fri Dec 30, 2016 2:44 pm**

Through games played on 12/29

Projected playoff matchups:

GSW vs DEN
HOU vs OKC
SAS vs MEM
LAC vs UTA

TOR vs WAS
CLE vs CHI
BOS vs DET
MIL vs CHA

Unweighted:

Code: Select all

Team  Rating Exp Wins
GSW	10.34  66.33
TOR	6.63	58.72
SAS	6.09	57.45
HOU	6.03	57.30
CLE	5.59	56.23
LAC	5.03	54.83
UTA	3.18	49.96
BOS	2.04	46.78
MIL	1.76	46.00
CHA	1.51	45.31
DET	0.42	42.20
OKC	0.27	41.76
MEM	0.26	41.73
CHI	0.06	41.17
DEN	-1.38	37.05
WAS	-1.40	37.00
MIN	-1.52	36.67
SAC	-1.54	36.60
ATL	-1.77	35.97
NYK	-1.91	35.56
POR	-2.14	34.91
IND	-2.19	34.77
MIA	-2.58	33.70
NOP	-2.84	32.98
DAL	-3.23	31.91
ORL	-4.08	29.63
LAL	-4.45	28.65
PHX	-5.15	26.88
BRK	-5.94	24.92
PHI	-7.06	22.30

Weighted:

Code: Select all

Team  Rating Exp Wins
GSW	10.25  66.17
TOR	6.48	58.37
HOU	6.08	57.41
SAS	5.98	57.17
CLE	5.41	55.79
LAC	4.45	53.33
UTA	3.05	49.60
BOS	2.13	47.06
MIL	2.00	46.70
CHA	1.45	45.13
MEM	0.46	42.31
OKC	0.17	41.48
DET	0.16	41.45
CHI	-0.28	40.19
WAS	-1.21	37.54
DEN	-1.37	37.08
SAC	-1.49	36.76
MIN	-1.54	36.61
NYK	-1.80	35.87
IND	-2.01	35.28
ATL	-2.05	35.18
POR	-2.12	34.98
MIA	-2.69	33.38
NOP	-2.71	33.34
DAL	-3.03	32.46
ORL	-3.78	30.42
LAL	-4.62	28.23
PHX	-4.96	27.34
BRK	-5.72	25.44
PHI	-6.66	23.19

Posted: **Sat Dec 31, 2016 5:42 am**

Median SRS per B-R is -1.58 at the moment. If that holds it would be the 2nd lowest value for a season in the history of the NBA. Seems the league is quite watered down at the moment. They'll never disband a franchise because they're too greedy, but they should get rid of at least 2 teams.

Posted: **Sat Dec 31, 2016 5:55 pm**

How can you determine the 'watered down' nature of the league by the median of anything?
What I see is that the #15 Wiz are -0.87, and #16 Wolves are -1.25; the median is in between those.
The avg (mean) is of course zero.

There's not a bell shaped distribution curve.

Code: Select all

SRS   #   -#
8+    1    0
6-8   4    3
4-6   2    2
2-4   0    8
0-2   6    4

Right column is # of teams with SRS <0
You could eliminate a team every year, and you'd still have below-avg teams, and odd distributions.

It may be that starter-quality players are languishing on some over-stocked teams. David West getting 11 mpg for GSW; a couple of the Spurs maybe?
I notice your projections sum to < 41 wins per team.

Posted: **Sun Jan 01, 2017 8:05 pm**

Well it improved slightly since I posted it, but the median SRS at the moment would still be 3rd worst in the history of the league. I'm not an expert, but I don't think it matters if the distribution is normal or not for the median to have relevance. When 50% of the teams are rated at -1.12 as they are at the moment, (approximately a 38 win team), in my mind that's a watered down league in terms of how the talent is distributed across teams in this particular season. Obviously I don't think the below average teams this season are worse than teams from the 50's or 60's. Many of the below average teams of today would beat the best teams from the early eras with relative ease. I suppose the better term to use would perhaps be competitive balance. The Warriors and a few other teams are playing very well and everyone else is marginally above average or below average.

Yeah, both versions are around 1225. So it's only 0.167 wins per team, not enough to really substantially change the projections. If it was only 1200 or something much lower I'd be more concerned about the quality of the ratings. Just doing this as a hobby, not for gambling purposes so I can live with it being less than perfect for now.

Posted: **Mon Jan 02, 2017 1:38 pm**

If you just got rid of the 2 worst teams, it likely has the greatest effect on those middling teams. If the Hawks don't have the Sixers and Nets to beat up on, they miss out on easy wins, and they likely do worse, not better.

If the talent from the disbanded teams goes to the remaining teams in a dispersal draft, then it could be set up to most benefit those at the bottom. But if the 'problem' is really talent concentration in the elite teams, it's not really a solution to cannibalize teams at the bottom.

I see Brk is given a 41% chance to beat Utah, NOP are 25% likely to win at Cle, Den is 21% at GSW, Was 28% at Hou, etc. This doesn't seem too disparate.

Posted: **Mon Jan 02, 2017 2:35 pm**

Fair enough. I suppose we'll have to agree to disagree. You make some good points. I guess a better solution to the problem would be to institute a hard cap like the NFL rather than eliminate teams. That would make it much more difficult, if not impossible, to create super teams like the Heat did with the Big 3 or Golden State adding Durant.

Where did you get those win probabilities from?

According to 538, BRK is 24% to beat UTA, NOP is 14% to beat CLE, DEN is 7% to beat GSW, and WAS is 16% to beat HOU.

For the games that have a moneyline at Pinnacle, the implied odds are:

BRK 27.88% to beat UTA and DEN 8% to beat GSW

Posted: **Mon Jan 02, 2017 2:49 pm**

Those probabilities appear here
http://www.basketball-reference.com/
... under Upcoming Games at the right.
Some interesting differences for sure.

Number of teams above or below median in ...

Code: Select all

stat:  MOV  SOS  SRS  ORt  DRt  Pace
avg=   .00  .00  .00  108  108  96.2
above   13   15   13   14   15   17
below   17   15   17   16   15   13

stat: FTr  3PAr TS%  eFG%  TO%  ORb%
avg=  .27  .31  .55  .51  12.9  23.5
above  16   11   12   11   12    15
below  14   19   18   19   18    15

Besides SRS, we see better shooting (and more 3fga) and higher TO concentrated among few teams.
Slower pace also is skewed down by a relative few.

Posted: **Mon Jan 02, 2017 3:59 pm**

Thanks for the link. I wonder how they're calculating those probabilities because they seem to be a bit off. Just going off current SRS ratings and adding 2.5 points for HCA, these are the hypothetical spreads for each game:

UTA -9.22
CLE -12.95
GSW -16.73
HOU -10.83

Using the chart at the link posted below, that would roughly translate to the following win probabilities:

UTA 82%
CLE 89%
GSW 97%
HOU 87%

If I use a regressed version of SRS (add 7 games of 0 to each team's rating), I get the following win probabilities:

UTA 74%
CLE 83%
GSW 92%
HOU 84%

Current win probabilities based on Pinnacle moneylines:

UTA 70.51%
CLE 81.47%
GSW 91.88%
HOU 80.5%

So the B-R projections are overvaluing the underdog's chances in each game by an average of 9.84% compared to Pinnacle's ML. Considering Pinnacle is taking 5K bets right now on the ML and allowing you to rebet once the line moves, I don't think the lines are off by that much. Otherwise someone like Haralabob would be hammering those lines and causing the odds to shift to be more in line with the B-R projections, if those were the true odds.

Convert spread to win probability: https://www.teamrankings.com/nba/odds-history/results/

Posted: **Mon Jan 02, 2017 4:30 pm**

Over at the b-r.com projections page --
http://www.basketball-reference.com/fri ... f_prob.cgi
-- you can see that projected Remain-ing W-L records are heavily regressed toward .500, relative to Current marks.

For example, the .853 GSW are projected to go .750 (36-12) the rest of the way, to finish at .798

I had assumed they were allowing for late season tanking, resting starters, and other uncertainties. But they're apparently assuming regressions start right now, as though games to date are not full indicators of team strength.

Posted: **Mon Jan 02, 2017 8:54 pm**

Thanks for the link. Either they are regressing teams too heavily or their formula for predicting win probabilities is skewed too heavily towards underdogs. There's no way Pinnacle's lines are off that much for NBA moneylines.

I updated my regressed SRS win probabilities using the odds history from TeamRankings and those are significantly closer to the Pinnacle win probabilities than the B-R ones.

Posted: **Thu Jan 05, 2017 6:35 pm**

In the last 30 NBA seasons -- since 1988 -- there have been 50 teams with a Win% < .240
http://bkref.com/tiny/WaWTV
That's equivalent to <20 wins in a season. This season, no team is below that mark, and no team is expected to win fewer than 24 games.

A 24-58 record is .293 W%
Last year with no team <.293 was 1984
The Nets are ranked worst team by SRS at -7.85
Other years the worst team was not as bad: 2007, 2004, 1992, 1984-86...

Posted: **Fri Jan 06, 2017 3:20 am**

There's hardly been any difference between the unweighted and weighted version, so will just be posting the unweighted one for the time being.

Sixers have the worst rating despite having the easiest SoS in the league with an avg opponent rating of -0.88 which equates to roughly a 38 win team.

Dallas (1.64) and Portland (1.36) are the only teams whose SoS is greater than 1. The avg opponent for each of them has been a 44-45 win team.

SoS ratings factor in opponent's faced and location (i.e. playing the Warriors on the road helps your schedule rating more than playing them at home). The B-R SoS ratings don't factor in location as far as I'm aware.

Code: Select all

Team  Rating Exp Wins
GSW	9.93	65.60
SAS	6.66	58.80
HOU	6.48	58.36
TOR	5.98	57.19
CLE	5.23	55.32
LAC	4.01	52.20
UTA	2.96	49.36
BOS	2.13	47.06
MIL	1.79	46.09
CHA	1.31	44.75
OKC	0.73	43.10
DET	0.21	41.60
MEM	0.10	41.30
CHI	0.00	40.99
ATL	-0.98	38.20
WAS	-1.17	37.65
MIN	-1.45	36.85
IND	-1.56	36.55
DEN	-1.75	36.02
SAC	-1.83	35.79
POR	-1.97	35.41
NYK	-2.45	34.05
NOP	-2.51	33.89
MIA	-2.66	33.46
DAL	-2.89	32.83
LAL	-4.11	29.54
ORL	-4.31	29.03
PHX	-4.74	27.91
BRK	-6.48	23.62
PHI	-6.66	23.21

Posted: **Thu Jan 19, 2017 7:56 pm**

Here's a link to a webpage published from the Google sheet where I track the ratings. The weighted version is the one being published at this link since that should be more relevant going forward in case of injuries, trades, etc.

https://docs.google.com/spreadsheets/d/ ... ingle=true

Houston currently on pace to exceed their Vegas win total by ~15.

Luckiest team is Cleveland whose actual win percentage exceeds their actual win percentage by 11%. A factor of them having a 0.875 win percentage in close games.

Unluckiest team is Minnesota whose expected win percentage exceeds their actual win percentage by 14%. A factor of them having a 0.231 win percentage in close games.

Close games winning percentage found here:

https://www.teamrankings.com/nba/stat/w ... lose-games

Playoff matchups at the moment:

TOR vs DET
CLE vs ATL
BOS vs CHA
MIL vs WAS

GSW vs MIN
SAS vs OKC
HOU vs MEM
LAC vs UTA

Finals projection at the moment is GSW vs TOR due to CLE's recent swoon, but I think they will overtake TOR by the end of the season and I would likely never bet on TOR to beat CLE in a best of 7 even if they had HCA over CLE. The best 3 teams are in the West and 7 of the top 10 are in the West. 7 of the teams in the 11-20 range are in the East and the bottom ten teams are split equally.

Based on current expected win totals, GSW would be healthy favorite in the finals. According to the tool linked below, they would have an 83.25% chance of winning a best of 7 series versus TOR assuming GSW had HCA. They would likely be around -500 to win the series at Pinnacle with that probability.

http://www.hardballtimes.com/tools-game ... abilities/

APBRmetrics

2016-17 Power Ratings

Re: 2016-17 Power Ratings

Re: 2016-17 Power Ratings

Re: 2016-17 Power Ratings

Re: 2016-17 Power Ratings

Re: 2016-17 Power Ratings

Re: 2016-17 Power Ratings

Re: 2016-17 Power Ratings

Re: 2016-17 Power Ratings

Re: 2016-17 Power Ratings

Re: 2016-17 Power Ratings

Re: 2016-17 Power Ratings

Re: 2016-17 Power Ratings

Re: 2016-17 Power Ratings

Re: 2016-17 Power Ratings

Re: 2016-17 Power Ratings