Wednesday, April 13, 2011

The Basketball I.Q. 2010-11 NBA Most Valuable Player Award

In this week’s post, I will determine the players most deserving of the National Basketball Association’s Most Valuable Player award for the 2010-11 season.  To make such an analysis more accessible, I will start with a glossary of statistical terms, which can be referred to by the reader:
SPR = [2PFGM + 1.5(3PFGM) + (FTM/2) + AST]/[FGA + (FTA/2) + AST + TOV]
TOT = TOV/[TOV + FGA + (FTA/2) + TRB + STL + AST]
SAR = [FGA + (FTA/2)]/AST
E = SPR + TAPPS + (1 – TOT)
wCE = (MPG/48) x [SPR + TAPPS + (1 – TOT)]
P/E = Salary/[SPR + TAPPS + (1 – TOT)]
wP/E = Salary/(MPG/48) x [SPR + TAPPS + (1 – TOT)]
EG = (Present Year’s E – Previous Year’s E)/(Previous Year’s E)
wEG = (Present Year’s wCE – Previous Year’s wCE)/(Previous Year’s wCE)

In writing this post, I set out to utilize the Basketball I.Q. advanced metrics to determine this season’s NBA Most Valuable Player.  This stands as Basketball I.Q.’s inaugural effort to determine the league MVP and, as I think the readers will see, the advanced metrics lend a fair and objective assessment to the evaluation of the NBA’s elite players.

However, it became clear to me as I calculated the stats that there are really three different ways in which one might determine an MVP:  In one instance, for example, you might try to determine the player who was truly most valuable – in other words, the player who gave you the most performance for the dollar, and thus the player whose contract represented the greatest value.

In another view, however, one could throw the acquisition cost of a player out the window and just look at what he does as a basketball player, independent of how much money that performance might cost his team in dollars.  But even this, I realized, could be viewed in two different manners:  Since players, even the league’s elite, play very different average minutes per game, this view of MVP – i.e., the best player, no matter the cost – could also be parsed by two different prisms.

In one, for example, the minutes played per game are totally ignored, and thus what you are truly measuring is how good a player is when he is on the court – in other words, who is the best player when his laces are tied, his sweats are off, and he’s in the game.  But in an alternative view, recognizing that different players might play very different amounts of minutes, you might recognize the weight of total minutes played, and correct for that – and in this instance, what you are measuring is the player who is the best over the course of 48 minutes, recognizing not just what he does when he is on the court, but for how long he does it.

Not surprisingly, my analysis yielded three different MVP’s:  One who, dollar for dollar, is the best player in the league; one who, when he is on the court, is unsurpassed in his execution; and one who, when his extensive minutes logged are considered, gives his team more minutes of excellent performance than anybody else.

I chose nine finalists for the Basketball I.Q. MVP award, all of whom are on playoff teams.  In alphabetical order, they are: Lamarcus Aldridge, Portland; Kobe Bryant, Los Angeles Lakers; Kevin Durant, Oklahoma City; Dwight Howard, Orlando; Lebron James, Miami; Dirk Nowitzki, Dallas; Chris Paul, New Orleans; Derrick Rose, Chicago; and Dwyane Wade, Miami.

The traditional basketball statistics are published in numerous venues, and so will not be reviewed here.  We will begin with one qualitative statistic (the Shot-to-Assist Ratio, or SAR) and three quantitative statistics (Successful Possession Rate, SPR; Turnover-Adjusted Points per Shot, TAPPS; and Turnovers per Touch, TOT).

                        SAR     SPR      TAPPS   TOT

Aldridge          9.64     .537     .985       .056
Bryant             4.96     .556     .954       .079
Durant             8.75     .571     1.032     .072
Howard           14.58   .530     1.005     .089
James              3.27     .607     1.006     .084
Nowitzki          7.28     .600     1.099     .060
Paul                 1.42     .684     .983       .068
Rose                2.96     .591     .941       .087
Wade               4.98     .573     .996       .082

Some things worth noting before we accumulate the quantitative statistics into valuations:  With the lone SAR in the lowest quintile (<2.89), Chris Paul is the only player in the MVP discussion who is a Primary Distributor (what the traditionalists would call a point guard).  Every other quintile has two representatives vying for the MVP honor: Derrick Rose and Lebron James are the only Combination Distributors (SAR between 2.89 and 4.70, or what the traditionalists might call a shooting guard);  Kobe Bryant and Dwyane Wade are the only Balanced Scorers (SAR between 4.71 and 6.75, or what traditionalists might call a small forward); Kevin Durant and Dirk Nowtizki are the two Primary Finishers (SAR between 6.76 and 9.21); and Lamarcus Aldrige and Dwight Howard are the two representative Absolute Finishers (SAR above 9.22, or what is often referred to as a center).

Looking at the quantitative statistics before converting them into value, a few things stand out:  For a player who is constantly dealing the ball around the court, Paul hardly ever commits a turnover -- this is a remarkable achievement.  James, Nowitzki, Durant and Howard shoot with tremendous efficiency – a statistic even more remarkable for Howard, oddly enough, because he commits an uncommonly large number of turnovers for a big man, and shoots his free throws so poorly (both of these work against his TAPPS, and yet his is still impressive).  And in the discussion of who is more important to the Miami Heat, James or Wade, it is worth noting that James is more of a distributor than Wade (as their respective SARs attest), and yet James still scores points with greater efficiency (as evidenced by their TAPPS) -- which is not common for a player who looks to pass as much as James does.  In fact, James turns the ball over only slightly more than Wade, who is less willing to pass the ball (usually, willing passers commit more turnovers).

The next table will evaluate each players Earnings (E, which corresponds to what a player does when he is on the court); weighted Cumulative Earnings (wCE, which accounts for playing time, and is thus a better measure of what a player does over a full 48 minutes, including his time on the bench); and weighted Price/Earnings ratio, which adjusts a player’s value for his salary.

                        Salary($M)      E          wCE     wP/E

Aldridge          10.70               2.47     2.05     5.22
Bryant             24.81               2.43     1.71     14.51
Durant             6.05                 2.53     2.06     2.94
Howard           16.51               2.45     1.93     8.55
James              14.50               2.53     2.04     7.11
Nowitzki          17.23               2.64     1.89     9.12
Paul                 14.94               2.60     1.96     7.62
Rose                5.57                 2.45     1.91     2.92
Wade               14.00               2.49     1.93     7.25

So, let’s start with the player who is truly most valuable – the one who gives you the most performance for his salary.  To determine this, one would use the wP/E, which divides a player’s salary (his “price”) by his weighted earnings on the court (in this manner, the lower the wP/E, the greater the value the player brings you).  As such, the first ever Basketball I.Q. Most Valuable Players, listed one to three, are:

1.      Derrick Rose (2.92)
2.      Kevin Durant (2.94)
3.      Lamarcus Aldridge (5.22)

In this case, Rose edges out Durant not because his performance is better (indeed, the advanced metrics suggest that Durant outplays Rose by a significant margin), but because his salary is smaller.  The same is true when comparing Rose to Aldridge:  Aldridge’s value over the course of 48 minutes appears to be greater than Rose’s (in large part because of the heavy minutes Aldridge logs), but he gets paid almost twice as much.  But this award isn’t looking for the best player, it is looking for the most valuable, and in 2010-11, the clear winner (though by a narrow margin) is Derrick Rose.

But now let’s look at who is the best player when he’s checked into the game and taking it at the opponent.  The Earnings statistic (E) would be the best measure of this, since it accumulates the additive value of SPR, TAPPS and TOT without regard for playing time or salary.  As such, the first ever Basketball I.Q. Best Guys on the Court, listed one to three, are:

1.      Dirk Nowitzki (2.64)
2.      Chris Paul (2.60)
3.      Kevin Durant (2.53)
Lebron James (2.53)

In this case, salary is not considered, nor are minutes played, and Nowitzki – with his remarkable scoring efficiency (TAPPS), an excellent SPR buffeted by high percentage shooting from the field and the foul line, and an uncanny ability to take care of the ball (a very low TOT) – seems to be the most “go-to” of all the go-to guys.

Finally, let’s look at the statistic that weights performance by playing time, the wCE.  This statistic, in essence, tells you who does the most over the course of a whole game, factoring in how much time he spends on the bench (and thus is unable to provide his team the benefit of his excellent skills).  As such, the first ever Basketball I.Q. Best Players for a Whole Game, listed one to three, are:

1.      Kevin Durant (2.06)
2.      Lamarcus Aldridge (2.05)
3.      Lebron James (2.04)

This award – which probably tells you who really is the most valuable player to his team, independent of salary – is remarkably tight: Durant, Aldridge and James all accumulate a ton of minutes, and, statistically, sit right on top of one another.  Of the three “awards” – MVP, Best Guy on the Court, Best Player for the Whole Game – I think the last is the most important determinant of game achievement, while the first (MVP, considering monetary value) is the most important factor for any GM trying to put together a team -- particularly in the dawning era of a hard salary cap.

There is one final note on which I would like to comment:  Of all nine players on this list, only one made the top three for Most Valuable Player, Best Guy on the Court, and Best Player for a Whole Game.  Aldridge and James each made two out of three, which speaks to what wonderful seasons they had, and how important they are to their teams -- but they did not crack all three lists.

Only Kevin Durant brought it whenever he was on the court; brought himself onto the court whenever he possibly could; and charged his team relatively little for the outstanding services he provided.  And so, for the unified crown of MVP, the choice is clear:

Kevin Durant, Oklahoma City Thunder.

Monday, April 4, 2011

Shelvin Mack and the NBA Draft: Why Mack Should Stay in College

In this week’s post, I will advance the argument that one of my favorite players in college basketball – Shelvin Mack of Butler University – should forgo the NBA for one more year, and return to college for his senior year of eligibility.  To make such an analysis more accessible, I will start with a glossary of statistical terms, which can be referred to by the reader:
SPR = [2PFGM + 1.5(3PFGM) + (FTM/2) + AST]/[FGA + (FTA/2) + AST + TOV]
TOT = TOV/[TOV + FGA + (FTA/2) + TRB + STL + AST]
SAR = [FGA + (FTA/2)]/AST
E = SPR + TAPPS + (1 – TOT)
wCE = (MPG/48) x [SPR + TAPPS + (1 – TOT)]
P/E = Salary/[SPR + TAPPS + (1 – TOT)]
wP/E = Salary/(MPG/48) x [SPR + TAPPS + (1 – TOT)]
EG = (Present Year’s E – Previous Year’s E)/(Previous Year’s E)
wEG = (Present Year’s wCE – Previous Year’s wCE)/(Previous Year’s wCE)

I am very much looking forward to tonight’s game for the NCAA Men’s Division I basketball championship.  Both teams, Butler University and the University of Connecticut, appeal to me for a common reason – they both play an intense brand of team basketball on both ends of the floor – as well as for differences in the most obvious aspects of their profiles:  Butler is clearly an underdog, a small school whose team system has cultivated an identity greater than any individual on their team (including its coach); meanwhile UConn is so compelling to watch because of both the wonderful exploits of one player in particular (Kemba Walker) and, love him or hate him, a coach (Jim Calhoun) who stridently preaches the virtues of a proper team game – while simultaneously, perhaps ironically, displaying a personality that is bigger than the team itself.

But the focus of today’s post is going to be Shelvin Mack, the guard for Butler University who, over the last two post-seasons, has played himself into the national spotlight – despite the fact that, like all underdogs, he arrives from a modest pedigree.  My concern for Mack, as wonderful as I think he is (and as much as I hope to see him get as much out of a basketball career as he desires), is that he might believe all the hype, and forgo his final year of college eligibility in an effort to turn professional.  As recently as this morning, on as big a platform as the New York Times, Mack has been hailed as a fringe first-round pick in the upcoming NBA draft – and though this may be true, I believe Mack can do even better if he returns to college, works on his game, and puts off the professional world until 2012.

As a basis of comparison, and to better illustrate why I believe Mack should return to school to polish his game, I will compare his statistics (both the traditional and alternative) over the last two years with those of his counterpart in tonight’s championship game, Kemba Walker of UConn, who is undoubtedly ready for the professional ranks, and will likely be a top 10 pick in the draft.  We will start with the traditional statistics:

                        PPG      AST      REB      FG%     FT%      3P%

Mack, 2010     14.1     3.0       3.7       .454     .734     .391
Mack, 2011     16.1     3.5       4.4       .412     .773     .353

Walker, 2010  14.6     5.1       4.3       .403     .767     .339
Walker, 2011  23.7     4.6       5.4       .432     .818     .336

If you go by the traditional statistics, you will see that in 2010, the sophomore year for both players, Mack and Walker put up very similar numbers – Walker had slight edges in scoring, rebounding and free throw shooting, and a sizable advantage in assists; Mack had a big edge in field goal and three-point shooting.  If you were blinded to the identities of each player, and only had access to their statistics, it would be very hard to tell the sophomore Mack and the sophomore Walker apart.  Indeed, if all you knew about them was that they were sophomore guards playing at a nationally competitive level of Division I basketball, a reasonable person might conclude that each player was very good, but probably not ready for the professional ranks – which likely would have been true.

But there has been some separation between Walker and Mack during their junior years:  Most obviously, Walker has increased his scoring by over 60%, while Mack increased his by about 14%.  And though both players showed improvement in most statistical categories, Walker has seemed to improve more:  Both were pretty good free throw shooters last year who have gotten better this year, but Walker has widened the gap in his superiority; both got a fair amount of rebounds for backcourt players last year, and are doing even better this year, but Walker, again, has widened his considerable advantage; and although Walker’s assists have gone down this year (as he has become the finishing focus of the UConn offense), and Mack’s have gone up, Walker still averages more than one assist a game more than Mack.

What might be most telling about the traditional statistics and the professional readiness of these two players, however, is their field goal shooting:  As each player gained a year of experience, Walker demonstrated improvement in both overall and three-point shooting, while Mack declined in both.  In defense of Mack, he clearly benefited from the presence on his team last year of Gordon Hayward, who is now playing in the NBA, and has probably had to work harder to find his shots on the floor.  But as of the last game of their junior years, Walker is the slightly better field goal shooter, and he and Mack have similar skill behind the arc.

A reasonable person might review the statistics of these two players’ junior years, blind to their identity, and conclude that both are very good – but only one is clearly ready for the pros.  And that player would be Walker.

Now let’s turn our eyes to the alternative statistics, which I believe will argue the case, even more emphatically, that Mack is not quite ready for the NBA.  Again, we will look at Mack – with the NBA-ready Walker as a basis for comparison – over the last two years.  We will start with the alternative statistics that more aptly describe the style of each player’s game (the qualitative measures), before we move on to the statistics that more accurately reflect execution (the quantitative numbers).  We will start with the Shot-to-Assist Ratio (SAR), the Shot Selection Index (SSI), the 3-Point Rate (3PR), and the 3-Point Skew (3PS):

                        SAR      SSI        3PR      3PS

Mack, 2010     4.15     .258     .435     .375
Mack, 2011     4.27     .322     .492     .421

Walker, 2010  2.78     .536     .297     .250
Walker, 2011  4.75     .428     .310     .241

So what do these qualitative statistics say about how each player approaches the game?  Well, for one, it would state that during the current season, both Mack and Walker play the same “position”:  With SARs of 4.27 and 4.75, respectively, both players would qualify within the second quintile of the Basketball I.Q. classification, known as Combination Distributors.  Interestingly, last season Walker had a much lower SAR – low enough to qualify him within the first quintile, referred to as Primary Distributors – but through a combination of refining his game over the summer and being asked to fill a different role by his coaches, he, too, is now a Combination Distributor.  As such, it is entirely fair to compare Mack to a standard that is defined by Walker, since they have virtually identical patterns in which they choose to dominate the ball or distribute it.

From there on in, there are qualitative aspects to Mack’s game that leads one to wonder whether he is ready for the next step.  For starters, his SSI this year is only .322, which is actually an improvement over last season’s .258.  Put simply, this means that Mack takes about one-third as many free throws as field goal attempts, which is just about the average for an NBA player.  The problem is, most players who go on to the pros see a significant dip in their SSI from their college days – meaning that they are taking shots that have a lower probability of going in, in which their opponents are less likely to foul, or defend with too much aggression.  While it is good news that Mack’s SSI improved from his sophomore year to his junior year, in reality he went from being below average in his shot selection to simply average – with the next level of competition representing a significant obstacle to improvement in this regard.

By contrast, Walker’s SSI from his sophomore year to his junior season went from absolutely astounding (.536, a number you might see in a low-post player) to simply excellent (.436) – and it is entirely possible that this calculated free throw rate went down because Walker is such an excellent free throw shooter (81.8%) that his opponents were better off letting him shoot, rather than giving him two free shots.

The other red flag that does not bode well for Mack’s performance at the next level is the rate at which he shoots 3-point shots, and the bulk of his scoring for which those shots account.  Nearly half of Mack’s field goal attempts (.492) are 3-pointers, and it accounts for greater than 40% of his made baskets.  This is not necessarily a problem, except that at the next level, the three-point arc will be about three feet further back, and you would expect his college-based performance to thus fall off significantly in this regard.  Furthermore, Mack’s 3PR and 3PS – both of which increased significantly from his sophomore to junior season – might be of less concern if he were a remarkably good 3-point marksman.  Mack’s 3-point shooting, however, declined over the last year – perhaps a reflection, again, of not having Hayward to relieve perimeter pressure, but it’s not as if the perimeter defense in the NBA gets any easier (unless you are playing the Knicks).

Walker, by contrast, does not rely significantly on the 3-point shot – in both his sophomore and junior seasons, they accounted for less than a third of his shots, and only a quarter of his makes.  As such, the contribution that Walker’s game receives from the 3-point ball would be expected to be less affected by an arc that is three feet further away.

Now let’s turn our eyes to the alternative statistics that define quantitative execution: the Successful Possession Rate (SPR), the Turnover-Adjusted Points per Shot (TAPPS), and the Turnovers per Touch (TOT).

                        SPR      TAPPS    TOT

Mack, 2010     .571     .963       .091
Mack, 2011     .552     .924       .093

Walker, 2010  .561     .861       .103
Walker, 2011  .573     .983       .063

Two things immediately stand out when you observe Mack’s statistics: 1) For the current season, they are nowhere near as good as the standard for comparison (Walker); and 2) they dropped off from last season to this.  Even using his best of the last two seasons (2010), you might argue that Mack just isn’t quite there: though his SPR and TAPPS last year were pretty good, his TOT is not good at all – Mack must take better care of the ball as he moves forward.  Walker, meanwhile, has a pretty good SPR this year, an extremely good TAPPS, but an absolutely phenomenal TOT – meaning he should be able to take most of whatever the NBA defenses throw at him.

There is one more point to consider, though for academic purposes only:  As discussed, the use of traditional statistics would have a reasonable person conclude that last season (2010), Walker was a better player than Mack.  However, confining ourselves to the 2009-10 season alone, a review of the alternative statistics clearly demonstrates that, last season, Mack was the better player:  He had a higher SPR (despite playing in a higher quintile than Walker last year, as higher quintiles generally result in lower SPR), a much higher TAPPS, and a better TOT.

This point is not intended to argue that Mack should enter the NBA next season, because last year he was better than an NBA-ready player (Walker).  Rather, it is meant to demonstrate that last year, neither player was ready for the NBA – and that this year, one player has demonstrated sufficient growth to warrant entry to the next level (Walker), while the other has exhibited a developmental plateau (Mack).  In fact, the next set of alternative statistics will demonstrate the growth differentials between the two players that ultimately define their readiness for the NBA:  Earnings (E), weighted Cumulative Earnings (wCE), Earnings Growth (EG), and weighted Earnings Growth (wEG).

                        E          wCE     EG                    wEG

Mack, 2010     2.44     1.88     --                      --
Mack, 2011     2.38     1.90     -2.46%             1.06%

Walker            ,2010   2.32     2.05     --            --
Walker, 2011  2.49     2.35     7.33%              14.63%

As a cumulative measure of all alternative statistics, Mack just does not meet the benchmark set by Walker.  And if you are trying to predict growth in the future – particularly growth that might occur at a more formidable level – Mack has been flat, while Walker has been explosive.  Both players have been able to sustain their games while playing more minutes (which is why Mack’s weighted earnings growth is positive, despite a negative unweighted earnings growth).

Taken altogether, the alternative statistics yield one surprise (Shelvin Mack was a better player than Kemba Walker in 2009-10) and one obvious conclusion (only Walker is now deserving of a lottery pick).  As such, it is my recommendation – indeed, my hope – that Mack will return to Butler for one more year of school, and one more season of basketball.  Such a decision, the statistics have me believe, would benefit Mack; benefit Butler; and, in 2012, benefit the NBA.

I suppose it might be argued that, just because one player is not as good as another, that does not necessarily mean the lesser player should not declare himself for the draft.  My retort to that argument is that, if a player is leaving eligibility on the table, then it is incumbent upon him to be every bit as good as any other player at his position who might also enter the draft – otherwise he is selling himself tremendously short.  And if the counter to that argument is one of hardship (the “Take the Money” argument), I would reply that this is not relevant to a late first-round or early second round pick – one extra year of $1 million really is not going to change anybody’s life, even a poor person’s, as weird as that may sound (perhaps a future post might take on the decision analysis of entering the draft one year early, and what the differences in post-tax income amount to -- I bet it's pretty small).

And, of course, the NBA is filled with stories of players who came back for one more year, despite the virtual certainty of being a first-round pick, and got themselves a decade’s worth of return for just one year of college basketball (e.g., Andre Miller of Utah, 1998-99) – and many more stories of players who left college a season or two early, and probably lost out on the better part of a career (e.g., William Avery of Duke, 1999-2000).

If Mr. Mack should read this post, I want him to know that I write this not because I am keen on being his critic, but because I greatly admire his game and wish him the best.  I’ll be pulling for Butler tonight, and would like nothing more than for Mack’s constancy to be rewarded.  I even look forward to watching him play at Madison Square Garden one day, hopefully for the home team.

Just not until 2012.