NBA MVP Watch: An Objective Assessment

I have developed my own statistical formula for determining the league MVP, a formula I will simply call the MVP Rating (the formula is given at the end of the article).  Historically, my system has been more accurate for determining the MVP than several of the best known systems out there (and I know of no system that’s been more accurate—if anyone does, I’d love to hear about it).  I’ll compare it here to John Hollinger’s PER system, the NBA’s own “Efficiency” system, and the Win Shares system (as kept at Basketball-Reference.com).  I’ll go back over the last 33 full NBA seasons (the NBA started keeping track of turnovers in 1978, 33 seasons ago, giving us the full range of data that we have today).

So, 33 full seasons since 1978, and 33 MVPs.

                        Rankings of actual MVPs in various rating systems

                        1^st        2^nd        3^rd        4^th        5^th        Other (with lowest ranks)

*MVP Rat.          21        6          2          2          0            2 (8^th, 6^th)

WSh                  18        6          3          2          0            4 (19^th, 15^th, 10^th, 10^th)

PER                  14        7          4          4          0            4 (15^th, 12^th, 8^th, 7^th)

Efficiency           14        6          3          3          1            6 (18^th, 14^th, 12^th, 11^th, 7^th, 7^th)

* my rating system

WSh (Win Shares) and PER rankings can be checked at Basketball-Reference.com.

PER is John Hollinger’s Rating System

Efficiency is the NBA’s rating system.

We are now a little more than a third of the way into the 2010-11 season (419 of 1,230 games have been played, 34.1%).  I’ll list the leaders in MVP Rating as of now.  I know that people will have emotional responses (“No way player x should be ranked that low, and player y should be much higher.  You’re crazy, etc...”), but I simply say, look at the results above.  I think they indicate that my system provides a pretty reliable indicator.  Here are the top MVP candidates, then, after Tuesday night’s games.

OSP is Overall Statistical Production (see formula below).  %OSP is the percentage of the team’s OSP that a single player accounts for.  PjW is the team’s projected wins.  The simple formula is %OSP times PjW.  The full details of the formula are given at the end of the article.

Rk	Player	Tm	G	GM	OSP/g	%OSP	PjW	MVP Rat.
1	Deron Williams	UTA	29	0	31.5	24.1	57	13.61
2	Dirk Nowitzki	DAL	28	0	26.6	19.8	67	13.32
3	LeBron James	MIA	30	0	29.8	22.8	57	13.11
4	Pau Gasol	LAL	29	0	29.6	21.5	59	12.79
5	Russell Westbrook	OKC	29	0	28.8	22.4	57	12.64
6	Tony Parker	SAS	27	0	23.9	16.8	73	12.25
7	Manu Ginobili	SAS	27	0	23.6	16.6	73	12.12
8	Rajon Rondo	BOS	20	6	31.5	17.4	69	12.07
9	Chris Paul	NOH	28	0	31.0	25.7	47	12.04
10	Tim Duncan	SAS	27	0	23.4	16.4	73	11.97
11	Kevin Garnett	BOS	26	0	23.5	16.8	69	11.69
12	Derrick Rose	CHI	25	1	29.0	21.1	54	11.33
13	Jason Kidd	DAL	28	0	22.1	16.4	67	11.05
14	Dwyane Wade	MIA	29	1	26.0	19.2	57	11.04
15	Paul Pierce	BOS	26	0	21.9	15.7	69	10.92
16	Lamar Odom	LAL	29	0	24.8	18.1	59	10.73
17	Kobe Bryant	LAL	29	0	24.8	18.0	59	10.70
18	Paul Millsap	UTA	29	0	23.5	17.9	57	10.12
19	Al Horford	ATL	30	0	25.9	20.5	49	10.10
20	Dwight Howard	ORL	26	2	27.1	21.4	47	10.01
21	Kevin Durant	OKC	25	4	25.9	17.4	57	9.82
22	Raymond Felton	NYK	28	0	27.8	20.5	47	9.62
23	Amare Stoudemire	NYK	28	0	27.7	20.4	47	9.58
24	Ray Allen	BOS	26	0	18.9	13.6	69	9.43
25	Joakim Noah	CHI	24	2	24.5	17.2	54	9.21
26	Josh Smith	ATL	30	0	23.1	18.4	49	9.03
27	Chris Bosh	MIA	30	0	20.6	15.7	57	9.03
28	Jason Terry	DAL	28	0	17.3	12.9	67	8.68
29	Al Jefferson	UTA	29	0	19.8	15.1	57	8.53
30	Carmelo Anthony	DEN	24	2	24.1	16.8	50	8.50

I don’t know how interested I am in arguing with people over these rankings—like I say, I think the results of the system over 33 seasons speak for themselves. 

People will say that Kobe is too low.  Well, welcome to the 2010-11 season (he’s not the Kobe of 2007-08—get used to it). 

Bulls fans will go nuts about Rose being only No. 12.  Sorry Chicago, you overrate him (though he is very very good, and could easily rise in the rankings in the months to come). 

Knicks fans—we all saw what happened when you were all crying out that Amar’e would be MVP. 

Durant fans—he’s off the MVP map, and Westbrook is the man right now, so get used to it (though Durant has been stellar lately, admittedly). 

Wade fans will go nuts to see LeBron ranked above him, and LeBron haters will object to his high ranking.  All I can say is, LeBron sweats production and wins, even when he’s not playing particularly well.  However, I do think that, given the negative response to “The Decision,” LeBron would have to have a huge lead, probably at least two above the second highest ranking, to be voted MVP. 

By the way, Kevin Love and Blake Griffin have high OSPs, but they’re not to be seen here because their teams project to so few wins.

Players within one of the leader’s MVP Rating can be considered to be seriously in the running.  Odds are reduced for players more than one but less than two off, and are further diminished for players more than two off.  Players that are more than 3.5 off the lead may be considered off the MVP map for now, though obviously two-thirds of the season remains to be played, and a lot can change.  I’ll point out that Chris Paul’s Hornets are in free fall, so expect him to go down week by week.  Rondo will keep slipping the longer he doesn’t play.  Rose is moving up, though not as rapidly as his supporters claim.  Durant has been on the rise, though I wouldn’t put money on his actually rising into the list of the contenders anytime soon.

How far off MVPs have been in my MVP rating, and chance of being MVP for players that far off in the future

Amount Off                            # (33 seasons)              Chance for future finishers

0 (1^st in MVP Rat.)                 21                                63.6%

Here is the full explanation of the formula.

MVP Rating = (OSP/TmOSP) * PjW

[OSP = Overall Statistical Production / Pj = Projected (to 82 games)]

OSP = Plays Made ± Surplus (or Deficit) Scoring ± Surplus (or Deficit) Possession Value

Plays Made = (Pts * 0.5) + Reb + Ast + Stl + Blk

Surplus (or Deficit) Scoring = (Pts – Expected Points)

Expected Points = (Scoring Attempts * (LgAvg Pts/Scoring Att))

            Scoring Attempts = ((FTA * 0.44) + FGA)

Surplus (or Deficit) Possession Value = ((Expected TO – TO) * (LgAvg Pts/Poss))

            Expected TO = (Ast * (LgAvg TO/Ast))

            Poss = (Scoring Attempts + TO – Offensive Rebounds)

            The current league average for Points/Scoring Attempt = 1.08

            The current league average for Points/Possession = 1.05

            Per game averages for ± scoring generally run from +3 to -3

            Per game averages for ± possession values generally run from +3 to -3, though Chris Paul and Rajon Rondo can get into the +4s