NBA MVP by the Numbers: An Attempt at Objectivity

Today I was reading yet another article on B/R proposing that the young rising superstar, Kevin Durant, should be this year’s MVP instead of LeBron James.  The writer used fairly reasonable criteria to laud Durant (apparently unaware that however stellar Durant is in each category, LeBron is ahead of him in all of them).

He considered: 1) stats; 2) importance to one’s team; and 3) the success of one’s team.  I was going to lay out the numbers on these categories as a post on that article, then realized the post would itself be article length, so I might as well write an article.

1. Overall Statistical Production (OSP)

I use my own formula for this.  I’m aware, of course, of Hollinger’s PER formula.  For reasons I won’t go into, I’m not crazy about it—I think it’s needlessly complicated, and I stick to Occam’s razor (basically, when two explanations both work, take the simpler one).

But I’ll just note that LeBron leads PER by far and apparently is likely to finish with one of the all-time high PERs, if not the highest.

The core of my formula is “plays made” (points times .5, plus rebounds, assists, steals, and blocks).

Why half the points?  Because a made field goal is two points but one “play made.”  I think that’s best, but it doesn’t matter—since LeBron is the leading scorer in the NBA, not halving points means he’s just further ahead of everyone else.

Of course, efficiency matters as well, but you have to figure out a way to combine percentages (like 50 percent shooting) with totals (like 20 points).

Here’s what I do: There is a league rate for points per scoring attempt (field goal attempts plus .44 times free throw attempts; you’d think it’d be .5 times free throw attempts, but it’s lowered to account for “and one’s”—.44 is the consensus number among stats geeks). 

I take the player’s scoring attempts and multiply by the league average of points per scoring attempt to get an “expected points” number.  If the player scores more than his expected points, this is added to his OSP.  If he scores less, it’s deducted.

The range is about +3 for the biggest efficient scorers in the league (LeBron leads in this category too, by the way), and about -2 for the worst offenders. 

Likewise, I come up with an “expected turnovers” number by multiplying the player’s assists times the league assist-to-turnover ratio.  If the player turns it over less than expected, the number is added to his OSP; and obviously if the player turns it over more, this is deducted from his OSP.

Again, the range goes from about +3 to -2.

So, to summarize, OSP (overall statistical production) is plays made, plus or minus a couple for efficient (or inefficient) scoring, plus or minus a couple for efficient (or inefficient) care of the ball (turnovers).

LeBron leads the league in “plays made” (by a lot), and he is second in scoring and ball handling efficiency points (behind only Steve Nash), so guess where he rates overall?

This chart shows the OSP leaders (as of Sunday morning, March 21):

Rk	Player	TM	G	OSP	OSP/G
1.	LeBron James	CLE	68	2639.4	38.8
	Chris Paul*	NOH	38	1333.5	35.1
2.	Deron Williams	UTA	64	1871.0	29.2
3.	Steve Nash	PHO	68	1975.4	29.0
4.	Dwyane Wade	MIA	66	1906.7	28.9
5.	Kevin Durant	OKC	67	1927.3	28.8
6.	David Lee	NYK	68	1935.9	28.5
7.	Chris Bosh	TOR	61	1720.4	28.2
8.	Dwight Howard	ORL	70	1960.8	28.0
9.	Rajon Rondo	BOS	68	1826.8	26.9
	Pau Gasol*	LAL	52	1389.4	26.7
10.	Carlos Boozer	UTA	67	1788.4	26.7
11.	Kobe Bryant	LAL	64	1701.8	26.6
12.	Jason Kidd	DAL	67	1756.0	26.2
13.	Dirk Nowitzki	DAL	68	1764.2	25.9
	Carmelo Anthony*	DEN	57	1476.0	25.9

*Paul, Gasol, and Anthony are not given rank numbers because they do not project to the NBA minimum of 70 games played (in order to be considered as having played a full season).

You can see that LeBron is way ahead of everyone except for Chris Paul, and Paul of course has not been playing for some time.  Durant is only fifth in the league.  It will probably shock people to see that Kobe is not in the top ten—“win shares” gives a better measure of his status in among the league’s elite, but we’ll get to that a little later.

I don’t know what to say to those who are surprised that Carmelo is not in the top ten, because it doesn’t surprise me.  He is basically a scorer; the people ahead of him score, but they make a lot other plays as well.

2.  Importance to One’s Team

I get accused of relying too much on the numbers, and not enough on just watching the games.  But I watch games every night, and I play basketball—quite competitively—three days a week.  Not that I can back it up in cyberspace, but experience tells me that even at 49 I’m probably a better player than, on a conservative estimate, 80 percent of those posting at B/R (and I’m trying to go as conservative as possible).

My point is not to toot my own horn (in a particularly empty way), but to say that I am passionately involved in the game beyond the numbers. 

Here’s what I like about numbers, though.  People will post saying that stats can lie.  I’d say that the wrong stats can be misleading (for instance, how can player A, with higher percentages in field goal shooting, three point shooting, and free throw shooting, be a less efficient scorer than player B?  Because player B gets to the free throw line at a much higher rate.)

Because the wrong stats can be misleading, you need to find the right stats.  That said, I would add the eye can lie more than the numbers.

When I watch Kenyon Martin slam home a putback, my eye tells me he must be one of the most dominating power forwards in the game.  The numbers tell me otherwise (that he is fundamentally just a role player).

When I used to watch Tracy McGrady shoot (in highlights), my eye told me he was a great shooter.  The numbers tell me otherwise (check them—his percentages were always low).

In any case, I don’t see why “importance to one’s team” can’t be quantified.  If you can come up with a number for a player’s OSP, you can get a number for the team’s OSP, and then it’s just one more step to get the percentage of the player’s contribution to his team (player OSP divided by team OSP).

Here are the league leaders:

Rk	Player	TM	G	Missed G	OSP	TmOSP	%OSP
1.	LeBron James	CLE	68	2	2639.4	9428.9	28.0
2.	Kevin Durant	OKC	67	0	1927.3	8452.1	22.8
3.	Dwyane Wade	MIA	66	4	1906.7	8373.0	22.8
4.	David Lee	NYK	68	2	1935.9	8648.4	22.4
5.	Dwight Howard	ORL	70	0	1960.8	9021.2	21.7
6.	Brook Lopez	NJN	69	0	1578.5	7336.7	21.5
7.	Steve Nash	PHO	68	1	1975.4	9702.3	20.4
8.	Rajon Rondo	BOS	68	1	1826.8	9018.0	20.3
9.	Gerald Wallace	CHA	64	5	1611.0	8121.0	19.8
10.	Zach Randolph	MEM	69	1	1721.6	8702.6	19.8
11.	Chris Bosh	TOR	61	7	1720.4	8782.8	19.6
12.	Josh Smith	ATL	68	0	1724.8	8883.1	19.4
13.	Andre Iguodala	PHI	70	0	1642.4	8482.6	19.4
14.	Dirk Nowitzki	DAL	68	1	1764.2	9180.2	19.2
15.	Jason Kidd	DAL	67	2	1756.0	9180.2	19.1
16.	Deron Williams	UTA	64	6	1871.0	9806.6	19.1
17.	Kobe Bryant	LAL	64	5	1701.8	9028.0	18.9
18.	Russell Westbrook	OKC	67	0	1579.7	8452.1	18.7
19.	Tim Duncan	SAS	64	3	1621.5	8739.2	18.6
20.	Carlos Boozer	UTA	67	3	1788.4	9806.6	18.2
21.	Marc Gasol	MEM	67	3	1568.5	8702.6	18.0
22.	Baron Davis	LAC	67	2	1493.8	8336.4	17.9
23.	Tyreke Evans	SAC	64	5	1481.9	8405.8	17.6
24.	David West	NOH	70	1	1505.1	8851.7	17.0
25.	Al Horford	ATL	68	0	1506.8	8883.1	17.0
26.	Andrew Bogut	MIL	62	6	1407.8	8308.5	16.9
27.	Al Jefferson	MIN	66	4	1355.8	8062.9	16.8
28.	Amare Stoudemire	PHO	69	0	1615.5	9702.3	16.7
29.	Joe Johnson	ATL	66	2	1457.7	8883.1	16.4
30.	Derrick Rose	CHI	65	4	1361.8	8302.8	16.4
31.	Nene Hilario	DEN	70	0	1523.7	9437.4	16.1
32.	Aaron Brooks	HOU	67	1	1323.2	8356.7	15.8
33.	Stephen Curry	GSW	67	2	1423.1	9005.0	15.8
34.	Carmelo Anthony	DEN	57	13	1476.0	9437.4	15.6

Obviously, Carmelo would be higher in this category if not for the 13 missed games; but he has missed 13 games, and the most games he could possibly play this season is 69.  No player has won the MVP with so few games played since the season went to 82 games, and it’s not going to happen with Carmelo this year.

As for the Durant fans out there claiming that he is more important to the Thunder than LeBron is to the Cavs, you can see that it’s not even close.  Durant is second in the league in importance to one’s team, but LeBron is a mountain above hills here. 

3.  (Credit for) Team Success

Obviously, it’s not just team wins we want here (Boobie Gibson, a role player on the league-leading Cavs, is not more valuable than Gerald Wallace, the best player on the Bobcats).

What we want is to know how much credit a player deserves for his team’s success.  Once we know his importance to his team, it’s an easy step to the amount of credit he deserves for their success: Simply apply his percentage of the team’s OSP times their wins.  This is the “win-share” concept.

Using my formula for OSP, below are the leaders.  Because the number of games different teams have played varies during the season, I rank them according to projected win shares (projected to 82 games).

Since a lot of people at B/R complain that wins are inflated in the “soft” Eastern Conference, I adjust projected win shares by strength of schedule (SOS).

Here are the numbers:

Rk	Player	TM	Tm Wins	%OSP	WinShr	Proj WinShr	SOS Adj
1.	LeBron James	CLE	55	28.0	15.40	18.04	17.67
2.	Dwight Howard	ORL	49	21.7	10.65	12.48	12.40
3.	Kevin Durant	OKC	42	22.8	9.58	11.72	11.74
4.	Kobe Bryant	LAL	51	18.9	9.61	11.43	11.56
5.	Rajon Rondo	BOS	45	20.3	9.12	10.83	10.57
6.	Steve Nash	PHO	43	20.4	8.75	10.40	10.57
7.	Dirk Nowitzki	DAL	46	19.2	8.84	10.51	10.53
8.	Jason Kidd	DAL	46	19.1	8.80	10.46	10.48
9.	Deron Williams	UTA	45	19.1	8.59	10.06	10.28