Sports Reference Blog

Is This the Best NBA MVP Race Ever? A Statistical Inquiry

Posted by Jonah Gardner on April 10, 2017

We're just a couple of weeks away from the most glorious period of the annual sports calendar, the NBA Playoffs. However, there's still one huge piece of unfinished business to settle before we get there: the race for NBA MVP. This year's race is going down to the wire as LeBron James, Kawhi Leonard, Russell Westbrook, and James Harden all remain in contention with just a few games remaining on the NBA schedule.

In the big picture, very little has changed since the last time we checked in on the race: Russell Westbrook is still averaging a triple-double, James Harden is still combining efficiency, volume, and playmaking to an unprecedented extent, LeBron James is still looking like Magic Johnson 2.0, and Kawhi Leonard remains the best two-way player in the league while dominating the ball and shouldering the offensive load in a way that no Spurs player has since Gregg Popovich took over as head coach.

In that blog post, I offered a relatively milquetoast opinion on the race. "Everyone is good and they all deserve to win" isn't the kind of scorching take you'll be hearing on Undisputed any time soon. But while hot takes on current events are a little outside both my comfort zone and the purview of this blog, hot takes on basketball history are another matter entirely. That's why, rather than addressing the question of who should win this year's MVP, I wanted to dig into the other question everyone's been asking: Is this the best MVP race of all-time?

To answer that, I went through every MVP race since the award was created in 1956, eliminating any season where the players who finished 1-4 in the MVP voting didn't also rank in the Top 400 all-time in Win Shares (WS).

You may have two questions at this point. If you're wondering why the sample was so large, skip on to the next paragraph. But if you're wondering what the heck a Win Share is, I'll give a brief explanation. Win Shares are an advanced basketball stat based on the baseball stat of the same name. In the basketball version, each team gets roughly one win share per win, which is then divided among the players on the roster according to how much that player contributed on offense and defense, using a formula detailed in the B-R glossary. According to Win Shares, Kareem Abdul-Jabbar holds the record for the best NBA season ever (1971-72) and the best career (273.4 WS).

While the Top 400 may seem like a large universe of player-seasons, it actually created a rather reasonably-sized sample for a few reasons. First, because WS are a cumulative stat, players who played in eras where they spent more time on the floor are at an advantage. As you may have heard, today's players are skipping more games, and playing fewer minutes in the games they do play, compared to older eras. So Top 400 is a large enough sample to span every generation.

And yet, MVP voters haven't always seen eye-to-eye with Win Shares. In fact, while 53 seasons since 1955 had a least four players make the Top 400, only 22 had four Top 400 players who also finished in the Top Four in MVP voting, counting this year and assuming that Kawhi, Russ, LeBron, and Harden finish, in some order, first through fourth. Once I had this somewhat reasonable sample of seasons, I added in the two lockout years (1998-99 and 2011-12), and calculated the standard deviation in Win Shares between the Top Four finishers in the MVP voting, eliminating every season with a standard deviation greater than two.

That narrowed it down to ten MVP races. I've used a few factors to rank them. In addition to Win Shares, I looked at Player Efficiency Rating and Value Over Replacement Player, advanced stats with somewhat different frameworks. I've tried to measure each race according to two factors: how many of the top four actually had a case to win and the strength of their collective performance. What makes this year so fascinating is that there are four players putting up historically great numbers, any of whom can win. So a year where there were four great seasons but one that clearly stands above the others would grade lower, as would a year with four very close player who were all just OK.

10. 2006-07 MVP Race:

1Dirk Nowitzki, Dallas Mavericks (1,138 vote points, 16.3 WS, 27.6 PER, 6.2 VORP)

2Steve Nash, Phoenix Suns (1,013 vote points, 12.6 WS, 23.8 PER, 3.9 VORP)

3Kobe Bryant, Los Angeles Lakers (521 vote points, 13 WS, 26.1 PER, 5.3 VORP)

4Tim Duncan, San Antonio Spurs (286 vote points, 13 WS, 26.1 PER, 6.3 VORP)

The Race, By the Numbers: Total WS: 54.9, Total VORP: 21.7, Average PER: 25.9, WS Standard Dev: 1.73

Coming in number 10 is the most recent race on our list, aside from 2016. At first glance, this looks like one where the voters got it right, but in digging into the number, I've come to believe that Duncan probably didn't get enough attention. He averaged 20.0 points per game, 10.6 rebounds per game, and 2.4 blocks per game; while Dirk had the edge in scoring (24.6 PPG), he lagged behind Big Fundamental in the other categories by both one and a half rebounds and one and a half blocks.

But there was more to this than mere stat-stuffing (as if that needs to be said when it comes to Timmy), as Duncan anchored the second best defense in the league that year and the Spurs had the best overall SRS, despite winning ten fewer games than Dallas. That being said, Dirk's Mavs still came in fifth in defensive rating, and traded spots with the Spurs when it came to offense (second versus San Antonio's fifth).

In the end, I'd still give the slight edge to Dirk for being the slightly more creative force, playing with slightly less support than Duncan (who had Manu Ginobili and Tony Parker on his team), and having a tiny advantage in True Shooting Percentage.

9. 1992-93 MVP Race:

1Charles Barkley, Phoenix Suns (835 vote points, 14.4 WS, 25.9 PER, 7.5 VORP)

2Hakeem Olajuwon, Houston Rockets (647 vote points, 15.8 WS, 27.3 PER, 8.5 VORP)

3Michael Jordan, Chicago Bulls (565 vote points, 17.2 WS, 29.7 PER, 8.9 VORP)

4Patrick Ewing, New York Knicks (359 vote points, 10.6 WS, 20.8 PER, 3.2 VORP)

The Race, By the Numbers: Total WS: 62.8, Total VORP: 32.5, Average PER: 27.3, WS Standard Dev: 1.16

I cheated a bit to get this race in here since Ewing's numbers are a little below the cutoff. However, I've seen it discussed in relation to this year's MVP race, and rightfully so, because it featured three of the best players in league history putting up monster seasons.

But the problem is that, unlike this year, there's a clear leader; he just happened to come in third. MJ led by every advanced metric, and he also put up scoring numbers that would make even Westbrook and Harden blush. MJ scored 32.6 points per game with a field goal percentage of 49.5%, leading both of this year's frontrunners.

And yet, that comparison makes for a good example of how far we've come in applying stats to these debates. While MJ had the edge in raw scoring, pace and time on the floor obscures it a bit. In fact, Westbrook is averaging 44.9 points per 100 possessions, more than MJ's mark of 43.0 in 1992-93. And while MJ outshot Harden from the field (49.5% vs 44.0%), it doesn't account for the fact that Harden has taken way nearly six more three-pointers per game than MJ, and making those are worth one point more than the two-pointers MJ was shooting instead. By effective field goal percentage, Harden actually was the more efficient shooter from the field (52.4% vs 51.5%).

8. 1998-99 MVP:

1Karl Malone, Utah Jazz (827 vote points, 9.6 WS, 25.6 PER, 4.1 VORP)

2Alonzo Mourning, Miami Heat (773 vote points, 7.9 WS, 24.6 PER, 3.4 VORP)

3Tim Duncan, San Antonio Spurs (740 vote points, 8.7 WS, 23.2 PER, 3.0 VORP)

4Allen Iverson, Philadelphia 76ers (319 vote points, 7.2 WS, 22.2 PER, 3.3 VORP)

The Race, By the Numbers: Total WS: 33.4, Total VORP: 13.8, Average PER: 23.9, WS Standard Dev: 1.03

This is another year that doesn't strictly adhere to our criteria, since the lockout deflated everyone's counting numbers. However, it was a very close race, and one where I think voters would have been justified in backing any of the top four. Iverson had the raw scoring lead, but his scoring and playmaking burden wasn't quite as heavy as it would get in his years as a legit MVP candidate. Without that, it's hard to make up for the shooting inefficiency, especially when compared to the three bigs.

While Malone has the lead in the advanced stats, my personal vote here would have gone to Mourning. 1998-99 was the year Mourning averaged a devastating 3.9 blocks per game to go along with a 20 and 11. And while Duncan had the edge in scoring and rebounding, the Heat were slightly better on offense and Mourning himself was a more efficient shooter.

7. 1982-83 MVP Race:

1Moses Malone, Philadelphia 76ers (724 vote points, 15.1 WS, 25.1 PER, 4.0 VORP)

2Larry Bird, Boston Celtics (364 vote points, 14.0 WS, 24.1 PER, 7.2 VORP)

3Magic Johnson, Los Angeles Lakers (304.5 vote points, 12.5 WS, 23.0 PER, 6.9 VORP)

4Sidney Moncrief, Milwaukee Bucks (226 vote points, 13.2 WS, 22.6 PER, 5.3 VORP)

The Race, By the Numbers: Total WS: 54.8, Total VORP: 23.4, Average PER: 23.7, WS Standard Dev: 1.12

Malone walked off with the award this year and so it doesn't seem like NBA-heads look back on it as fondly as other MVP races. But maybe they should, since it was actually closer than you might realize. In fact, VORP seems to prefer Bird.

I'd probably still take Malone given that, while Bird has the edge on assists and steals, Malone pulled down over 15 rebounds per game and blocked over two shots. Although Bird was better from the line and also took more than a few three-pointers, their True Shooting numbers are pretty close. However, for those who think "valuable" should be a reflection on where the team would be without him, it should be noted that both Malone and Magic had teammates finish in the top 10 (Julius Erving and Kareem Abdul-Jabbar, respectively), while Bird did not.

6. 1986-87 MVP Race:

1Magic Johnson, Los Angeles Lakers (733 vote points, 15.9 WS, 27.0 PER, 7.4 VORP)

2Michael Jordan, Chicago Bulls (449 vote points, 16.9 WS, 29.8 PER, 8.8 VORP)

3Larry Bird, Boston Celtics (271 vote points, 15.2 WS, 26.4 PER, 8.5 VORP)

4Kevin McHale, Boston Celtics (254 vote points, 14.8 WS, 24.0 PER, 5.6 VORP)

The Race, By the Numbers: Total WS: 62.8, Total VORP: 30.3, Average PER: 26.8, WS Standard Dev: 0.92

I think you can draw a sharp line between 7 and 6 on this list. While the bottom four races were plenty entertaining, these six are clearly in a class above. This one, for example, features three players who just about any NBA fan would put in the top 5-10 of all-time, all having absolutely dominant seasons.

Still, there's two reasons I'd rank this race below the top five. For one thing, there's basically no justification for voting for McHale ahead of Bird (even McHale's .600 FG% pales in comparison to Bird's shooting when you factor in what Bird did at the line and behind the arc).

The other thing is that I think MJ, once again, is probably a step above the others. The GOAT averaged 37.1 points, 5.2 rebounds, 4.6 assists, 2.9 steals, and 1.5 blocks in 1986-87, and you really owe it to yourself to go back and read those numbers a couple more times before continuing with this blog post.

The main knock on MJ's case is that the Bulls finished below .500, admittedly a big minus, but their SRS suggests they were actually the eighth best team in the NBA, with a point differential closer to a 43-39. I understand if you think that three wins don't make a huge difference, or prefer to measure this sort of thing by what actually happened on the floor, but, in my opinion, if you put up the only 37-PPG season since 1963, and have a TS% over .560, and also stuff every other column in the stat table to a nearly unprecedented degree, you're getting my MVP vote unless your team only wins, like, 20 games.

5. 1969-70 MVP Race:

1. Willis Reed, New York Knicks (498 vote points, 14.6 WS, 20.3 PER, N/A VORP)

2Jerry West, Los Angeles Lakers (457 vote points, 15.2 WS, 24.6 PER, N/A VORP)

3Kareem Abdul-Jabbar, Milwaukee Bucks (335 vote points, 13.8 WS, 22.5 PER, N/A VORP)

4Walt Frazier, New York Knicks (50 vote points, 15.0 WS, 21.1 PER, N/A VORP)

The Race, By the Numbers: Total WS: 58.6, Total VORP: N/A, Average PER: 22.1, WS Standard Dev: 0.62

This was, by Win Shares, the closest MVP race that qualified for the list. Just 1.4 WS separated West from Kareem, and the group of four had a standard deviation of 0.62 that only one other race, that we'll get to later, even approached. However, it's a little lacking in both total Win Shares (sixth out of the ten races) and average PER (dead last).

Unlike the McHale/Bird example, this one's interesting because it isn't super clear-cut which Knicks teammate had the edge. Though Reed won, Clyde had an edge in the advanced stats because he had superior passing numbers, a slightly more efficient shooting season, and some great rebounding for a player at his height and position. They were both great, as was The Logo who averaged over 31 points, but it's hard to overlook what Kareem did that year. As a rookie, he averaged 28 and 14 and chipped in four assists.

But defense is a factor as well, and Reed was the anchor on the league's best defense, with that year's Knicks posting a DRtg over four points better than the second best defense. We're talking about roughly the same gap in defense between this year's Spurs and Raptors, enough for me to admit the voters probably got this one right.

4. 2002-03 MVP Race:

1Tim Duncan, San Antonio Spurs (962 vote points, 16.5 WS, 26.9 PER, 7.6 VORP)

2Kevin Garnett, Minnesota Timberwolves (871 vote points, 15.6 WS, 26.4 PER, 9.0 VORP)

3Kobe Bryant, Los Angeles Lakers (496 vote points, 14.9 WS, 26.2 PER, 7.1 VORP)

4Tracy McGrady, Orlando Magic (421 vote points, 16.1 WS, 30.3 PER, 8.7 VORP)

The Race, By the Numbers: Total WS: 63.1, Total VORP: 32.4, Average PER: 27.5, WS Standard Dev: 0.69

This race seems to have been largely forgotten, as people reach for 1989-90 as the obvious point of comparison for this year's MVP race. However, this one poses a lot of the same questions in terms of what you prioritize in a basketball player and how you define valuable.

You can really think of this as two sub-races between the wings and the bigs. In Kobe and Tracy, you have two high-scoring wings who anchored their offense while chipping in excellent numbers across the board. Despite what you'd assume on reputation, it was actually T-Mac who was the superior scorer, while Kobe had the slight edge in rebounding and assists:

Most Valuable Player Table
Per Per Per Shoo Shoo Shoo
Rank Player Tm PTS TRB AST FG% 3P% FT%
3 Kobe Bryant LAL 30.0 6.9 5.9 .451 .383 .843
4 Tracy McGrady ORL 32.1 6.5 5.5 .457 .386 .793
Provided by View Original Table
Generated 4/6/2017.

But those numbers somewhat overstate Kobe's edge, because the Lakers played at a faster pace. Adjusted for that, it's basically a draw in rebounds and assists, and even clearer win in scoring for T-Mac:

Per 100 Poss Table
Rk Player TRB
3 Kobe Bryant 8.6 7.3 2.8 1.0 37.6
4 Tracy McGrady 8.5 7.2 2.2 1.0 42.0
Provided by View Original Table
Generated 4/6/2017.

Then there's the bigs. Duncan averaged 23 points per game while doing all the things you expect your defensive anchor to do. He pulled down 12.9 rebounds per game and blocked 2.9 shots, while leading the third best defense in the league. Garnett also averaged 23 and 13, but he also picked up six assists per game and chipped in over a steal. This season was Garnett at his peak unicorn, flashing the offensive skill and polish usually associated with shorter players while also performing the duties traditionally associated with the big man.

But the results weren't as good. The Wolves ranked 16th in defensive rating, a much wider gap than their two-spot lead in offensive rating over the Spurs (fifth vs seventh).

But that also may have been because the cast around Garnett was so lacking. The Spurs' net rating improved by 14.7 points per 100 possessions when Duncan was on the floor compared to when he sat, an excellent number that would rank behind only LeBron among this year's contenders. But the difference was even starker for KG, whose Wolves improved by 23.6 points per 100 when he was on the floor. They were a +6.1 team when he played and a -17.5 team when he sat, essentially the difference between this year's Rockets and a team that's two points worse than the lockout-year Charlotte Bobcats who set the record for worst winning percentage in a season.

3. 1962-63 MVP Race:

1Bill Russell, Boston Celtics (341 vote points, 13.5 WS, 18.2 PER, N/A VORP)

2Elgin Baylor, Los Angeles Lakers (221 vote points, 14.4 WS, 26.6 PER, N/A VORP)

3Oscar Robertson, Cincinnati Royals (191 vote points, 16.8 WS, 24.6 PER, N/A VORP)

4Bob Pettit, St. Louis Hawks (84 vote points, 14.3 WS, 25.0 PER, N/A VORP)

The Race, By the Numbers: Total WS: 59.0, Total VORP: N/A, Average PER: 23.6, WS Standard Dev: 1.42

This race may actually be the one that bears the greatest resemblence to this year's, because it poses a simple question to the voter: how do you define valuable? If you're looking for someone who made an undeniable impact in a way that showed up in every statistical category, you'd go with Oscar, who averaged 28-10-9 while also being the only one to shoot over 50%. Do you want a high-usage player shouldering his team's main creative burden to great success? Baylor took 28 shots per game for the team with the best record?

And yet, Russell won, despite relatively unimpressive numbers, because he was the focal point of a defense that held teams to a meager 87.4 points per 100. For those who would dismiss defense as a factor in the MVP race, it's worth noting that Russell's team had the best defense, the best record, and went on to win the NBA title in 1963.

2. 2016-17 MVP Race:

James Harden, Houston Rockets (14.8 WS, 27.3 PER, 8.8 VORP)

LeBron James, Cleveland Cavaliers (12.9 WS, 27.0 PER, 7.3 VORP)

Kawhi Leonard, San Antonio Spurs (13.4 WS, 27.6 PER, 6.1 VORP)

Russell Westbrook, Oklahoma City Thunder (13.2 WS, 30.8 PER, 12.4 VORP)

The Race, By the Numbers: Total WS: 54.3, Total VORP: 34.6, Average PER: 28.2, WS Standard Dev: 0.84

I'm very curious about where I'd rank this year's race if we were five or ten years removed, instead of right in the middle of it. Part of me wanted to rank it as number one; out of all of the races I looked at, this is still the most difficult hypothetical vote to cast. But is that just because I'm writing while the dust is still settling. Down the line, will there be someone as clearly deserving as, say, 1992-93 MJ was. Or will it still seem murky?

On total Win Shares, this is actually the least impressive field of any season on this list except for the lockout-shortened 1998-99. However, on average PER, it's the most impressive and on total VORP, it's second. But a lot of that is because Westbrook is putting up a season that's challenging for the record in both marks. By standard deviation for win shares, it's the fourth closest, but if I was using VORP instead, it'd be the least close.

In the end, each of these races poses a question. Some years, it's a question like "What do you want in a basketball player" or "Is offense more important than defense." I think this year's race is posing a more direct question: "What do you think of the year Russell Westbrook is having?" Do you think it's possible that someone can dominate the ball to the extent Russ is and have it be the right basketball strategy. Can an MVP come from a team that looks like this year's Thunder? How much should narratives and stories off the court, whether it's achieving certain round numbers or having to deal with the implosion of this generation's most fascinating partnership, factor into our calculation?

1. 1989-90 MVP Race:

1Magic Johnson, Los Angeles Lakers (636 vote points, 16.5 WS, 26.6 PER, 8.5 VORP)

2Charles Barkley, Philadelphia 76ers (614 vote points, 17.3 WS, 27.1 PER, 9.2 VORP)

3Michael Jordan, Chicago Bulls (564 vote points, 19.0 WS, 31.2 PER, 10.1 VORP)

4Karl Malone, Utah Jazz (214 vote points, 15.9 WS, 27.2 PER, 6.7 VORP)

The Race, By the Numbers: Total WS: 68.7, Total VORP: 34.5, Average PER: 28.0, WS Standard Dev: 1.35

This is the race that keeps coming up over and over again in comparison to this year's, and rightfully so. 1989-90 leads all the races on this list in both total Win Shares and total VORP, and that's without factoring in Patrick Ewing, David Robinson, and Hakeem Olajuwon, all of whom got first place votes but finished outside the Top Four. I originally had this lower because Malone seems like he's a step behind the others, but it's hard to fault a guy having a 30 & 10 season (the last, by the way, before Westbrook did it this year).

All four of these players averaged at least 20 points while posting a true shooting over 60%. Three of them averaged double-doubles, while the fourth averaged over 30 points per game. In the end, you can show this table to four different people:

Per Game Table
1 Charles Barkley* 1989-90 39.1 8.9 14.9 .600 .608 .749 11.5 3.9 1.9 0.6 3.1 25.2
2 Magic Johnson* 1989-90 37.2 6.9 14.4 .480 .526 .890 6.6 11.5 1.7 0.4 3.7 22.3
3 Michael Jordan* 1989-90 39.0 12.6 24.0 .526 .550 .848 6.9 6.3 2.8 0.7 3.0 33.6
4 Karl Malone* 1989-90 38.1 11.1 19.8 .562 .567 .762 11.1 2.8 1.5 0.6 3.7 31.0
Provided by View Original Table
Generated 4/6/2017.

And get four different answers about which season you'd rather build a team around. That's how you know it was a good MVP race. And, honestly, this year's may still be better.

5 Responses to “Is This the Best NBA MVP Race Ever? A Statistical Inquiry”

  1. Jeffrey Amos Says:

    I agree with 89-90 at the top. Controversial with Barkley receiving more 1st place votes than Magic.
    When measuring MVP seasons, I like to use a combination of PER (personal efficiency rating), WS (win shares), VORP (value over replacement player) and minutes played. Here's my top 20 for Apr.11

    1. Russell Westbrook
    2. James Harden
    3. LeBron James
    4. Giannis Antetokounmpo
    5. Jimmy Butler
    6. Kawhi Leonard
    7. Rudy Gobert
    8. Stephen Curry
    9. Karl-AnthonyTowns
    10. Isaiah Thomas
    11. Kevin Durant
    12. Anthony Davis
    13. Damian Lillard
    14. DeMarcus Cousins
    15. Chris Paul
    16. John Wall
    17. Kyle Lowry
    18. Nikola Jokic
    19. DeAndre Jordan
    20. Mike Conley

  2. Walt Coogan Says:

    Regarding the '02-'03 race, "adjusting for pace" constitutes a robotic fallacy. The relationship between pace and a star's statistics is not necessarily linear, and it may well be inverted. At a faster pace, the nature of the game changes. The ball is pushed up and around the court more quickly and thus no one player dominates the rock as much. There are more possessions, but a star does not dominate the possessions and occupy the basketball to the same extent. Conversely, at a slower pace, there are fewer team possessions, but a star is better able to milk and manipulate the possessions, to the benefit of his numbers. As the Lakers' pace slowed during Magic Johnson's career, did his numbers decline? Why did Chris Paul post his greatest statistics with the Hornets while playing at some of the slowest paces in the NBA? Why did Tony Parker's scoring peak before, rather than after, the Spurs' pace substantially quickened? Why did Michael Jordan and the Karl Malone/John Stockton tandem post historic numbers while playing at some of the NBA's slowest paces? Why did David Robinson win a scoring title with the slowest-paced team in the NBA?

    Frankly, PER constitutes a junk metric (largely, although not exclusively, for that reason), and Win Shares and VORP are fallacious or deeply problematic as well. Their results may sometimes make sense, but the reasons would be more coincidental or incidental rather than inherent.  

  3. Walt Coogan Says:

    The MVP constitutes a promotional gimmick, anyway, much like the Academy Awards. It epitomizes what historian Daniel Boorstin labeled a "pseudo-event," especially in a team sport where value is inherently context-dependent. 

    MVP or not, the best player in the NBA is LeBron James. 

  4. AC Says:

    I do not feel the #3.1962-63 NBA MVP race was a good example because did not include Wilt.

    Wilt Chamberlin, San Francisco Warriors, had significantly higher advanced stats than all top 4 vote getters, (9 vote Points, 20.9 WS, 31.8 PER) this was his highest PER of his career and he avg. 44.8 points on 52.8% shooting and 24.3 rebounds.

    1. Bill Russell, Boston Celtics (341 vote points, 13.5 WS, 18.2 PER, N/A VORP)

    2. Elgin Baylor, Los Angeles Lakers (221 vote points, 14.4 WS, 26.6 PER, N/A VORP)

    3. Oscar Robertson, Cincinnati Royals (191 vote points, 16.8 WS, 24.6 PER, N/A VORP)

    4. Bob Pettit, St. Louis Hawks (84 vote points, 14.3 WS, 25.0 PER, N/A VORP)

  5. Jeffrey Amos Says:

    I can't figure out why San Francisco wasn't able to win more games in 62-63. That's clearly the reason he's not included. However, he probably should in my opinion.