Given that the objective of NBA teams is to win (at least on the court), one would think that player evaluation would focus on that goal. When we look at the race for Rookie of the Year, though, it’s hard to see that focus.
The leading candidates for this award are Kevin Durant, Yi Jianlian, and Al Horford. Only one of these players, though, is playing for a team that has improved noticeably on last season’s results.
Last year the Sonics won 31 games with an efficiency differential (offensive efficiency minus defensive efficiency) of -3.0. This season the Sonics have a differential of -6.1. In sum, the Sonics with Durant are not as good.
The Milwaukee Bucks won 28 games last season with a differential of -4.6. This year the team’s differential is -5.2. In sum, the Bucks with Yi are not as good.
Turning to the Hawks, last season Atlanta won 30 games with a -5.1 differential. So far this season this team has a -0.5 differential. In sum, the Hawks with Horford are better.
My sense, though, is that both Yi and Durant are considered better rookies than Horford (at least, that’s how David Thorpe at ESPN sees it).
We could look at Wins Produced and see that this is not so, but let’s take a different approach. Kevin Durant replaced Ray Allen for the Sonics at shooting guard. Last year the Bucks employed a number of power forwards, but it appears Brian Skinner was the player most frequently employed at the position Yi currently occupies. And Horford has pretty much replaced Zaza Pachulia at center for the Hawks.
Imagine if these moves had not happened and the minutes currently allocated to the rookies were allocated to the players employed in 2006-07.
Part of the answer is revealed in Table One, where the per-48 minute performance of these six players – with respect to a variety of box score statistics — is compared to their position averages.
Table One: Durant, Yi, and Horford Again
Allen for Durant
Let’s start with Durant and Allen. When we look at the individual stats we see that Durant has been well below average with respect to shooting efficiency. And although he’s above average with respect to blocked shots and rebounds, with respect to net possessions (rebounds + steals – turnovers) he’s again below average. When we turn to Win Score per 48 minutes [WS48] we see a mark of 4.4. Yes, that’s well below average as well.
In contrast, Ray Allen was quite good. No, he wasn’t very good with respect to net possessions. But his skills as a scorer far outweighed these deficiencies. Consequently his WS48 was above what we see from an average shooting guard.
When we turn to WP48 we see the same story. Durant has a mark of -0.010 thus far this season, which is well below average. Allen had a WP48 of 0.144, which is above the average value of 0.100. To put these numbers in perspective, with Durant this season the Sonics projected Wins Produced is 24.4. If the Sonics simply replaced Durant with Allen, and Allen posted the same numbers we saw in 2006-07, the Sonics projected Wins Produced would be 33.1. In sum, virtually the entire decline we see in Seattle is tied to Durant taking over for Allen at shooting guard (yes, the other moves practically make up for the loss of Rashard Lewis). And this is the player people think is the best rookie?
Skinner for Yi
When we turn to Yi, we see a similar story, although not quite so dramatic. Yi has shown an ability to avoid turnovers and blocked shots. With respect to everything else he’s below average. Consequently his WS48 is only 8.9 while his WP48 is 0.020.
Although Yi is below par, he’s productivity is not far below what we saw from Skinner last year. When we look at the numbers we might think Skinner should be above average. He was above average with respect to shooting efficiency and rebounds. But because he didn’t take many shots (yes, it turns out you do have to shoot once in awhile), his WS48 and WP48 marks were below average. Still, if Skinner – with a WP48 of 0.039 — took the place of Yi this season, the Bucks would be slightly improved. Of course, one suspects that a rookie has to out-perform Skinner to be considered Rookie of the Year.
Pachulia for Horford
The Hawks actually could run this experiment for us. Pachulia is still employed by the Hawks, although his starting position has been taken by Horford.
When we look at the numbers we see that Horford is above average with respect to rebounds and shooting efficiency. Although he has problems with respect to blocked shots, assists, and free throw percentage, his overall WS48 and WP48 marks (WP48 of 0.171) are above the mean.
Except for blocked shots, Pachulia was almost the opposite player last year. He had trouble hitting his shots and was below average on the boards. Given these problems, overall his WS48 and WP48 marks (WP48 of 0.096) were slightly below average. In sum, going from Pachulia to Horford improves the Hawks.
Summarizing the Story
When we look at the scenarios outlined in this column, we see that Durant and Yi have made their teams win less. Horford has made his team win more. And since winning is the objective, how is the current debate about the Rookie of the Year really much of a debate?
There are two issues to consider. One, people might be focused strictly on scoring. By that measure, Horford lags behind Yi and Durant.
Although the RoY debate in 2008 might be another example of scoring being over-emphasized, I think it’s also possible that people are addressing a different question. Rather than look at “how productive has a player been?” people might be considering “how productive will a player be in the future?”
At times Durant looks like a great player. Of course, the data tells us that there are many, many times he’s not. We expect as time goes by, Durant will be “great” more often and perhaps, he might even be more productive than Horford. A similar story could be told for Yi. Such stories might be what people consider when they argue that Durant and Yi are better rookies than Horford.
At this point, though, these are just stories. All we know so far is that the Hawks are better, and part of that improvement is tied to Horford. As for the Bucks and Sonics, each team has declined with the addition of their high profile rookie. And for the Sonics, the decline is quite dramatic. Durant, so far, has simply not helped Seattle win games. And although he might someday, the story today is quite different from these dreams about tomorrow.
- DJ
Our research on the NBA was summarized HERE.
The Technical Notes at wagesofwins.com provides substantially more information on the published research behind Wins Produced and Win Score
Wins Produced, Win Score, and PAWSmin are also discussed in the following posts:
Simple Models of Player Performance
What Wins Produced Says and What It Does Not Say
Introducing PAWSmin — and a Defense of Box Score Statistics
Finally, A Guide to Evaluating Models contains useful hints on how to interpret and evaluate statistical models.
24 responses so far ↓
Panda Bear // December 31, 2007 at 1:27 am
This is an excellent post.
Brings up an interesting point. Is the “most valuable” rookie the rookie of the year?
Of course in practice the idiot sportswriters just pick whoever has highest ppg.
Pete23 // December 31, 2007 at 2:50 am
Great post. Durant might develop into a really good player, but some sportswriters are going overboard with the apotheosis of him.
Kent // December 31, 2007 at 4:25 am
Nice column!
Nick // December 31, 2007 at 6:26 am
Nice post, but do you think that this way of thinking penalises those rookies who are unfortunate enough to replace much better players? Yes Durant is worse than Allen would have been, but can Allen and Pachulia be compared really?
andrew // December 31, 2007 at 8:05 am
Nick,
Dave has already compared them strictly on their own merits and has shown Horford to be the superior rookie.
Today he decided to pursue a different tact (replacement value) and he still comes out on top.
-Andrew
Paulo // December 31, 2007 at 10:14 am
Dave,
Have you ever tried to correlate WP and Rookie of the Year/MVP voting? Not necessarily actual results, but maybe top 5 finishes?
Paulo // December 31, 2007 at 10:21 am
PS.
I realized it might be hard to find the previous voting results, but I would assume someone was diligent enough to find those.
dberri // December 31, 2007 at 10:32 am
Paulo,
We have looked at voting for the All-Rookie team by the coaches and performance. The same story we tell for salaries applies to voting for this award.
I am working on a paper looking at MVP voting. Again, it is still the same story. Scoring dominates all of these awards.
Jason // December 31, 2007 at 11:17 am
I suspect it’s more a story of sociology than economic value, but I find it fascinating how and when players get accepted as “stars” and what must happen to change this perception.
Certainly scoring 20ppg (even if it’s done more by hoisting shots than anything else) gets Durant attention, but I suspect that his “star” came long before he stepped out for the first jump ball. Durant was widely hyped before he ever suited up. People questioned whether he was going to be so good that passing on a potentially dominant center like Oden might have been a good move for Portland. This situation was compared to Jordan and Olajuwon.
It is possible for players, even high scorers, to be regarded as something less than stars. It happened to Glenn Robinson, who, by the end of his career was regarded as something of a ballhog who ‘didn’t make his teammates better’ and wasn’t ‘a winner.’ I wonder what sort of career Durant would have to take to be regarded similarly disregarded. Presently, it seems that his ‘potential’ outweighs his results, even when people acknowledge that his play so far hasn’t been good.
It is also possible for later picks to be regarded as stars (see Arenas and Boozer and, for a non-scorer, Ben Wallace) though in general, I suspect that for quite some time in the league, initial draft position influences opinion of a player as much as on-court production. What does someone have to do to override the evaluation that led to their draft position, either up or down?
dberri // December 31, 2007 at 11:27 am
Jason,
There has been research showing that draft position impacts minutes played well into a player’s career. And that is after controlling for performance.
magicmerl // December 31, 2007 at 11:50 am
Is the relationship between draft position and minutes correlation or causation do you think?
dberri // December 31, 2007 at 1:21 pm
merl,
Performance was controlled for. So this result tells us that teams are reluctant to abandon a choice. Even after they have learned that a player is not going to produce, minutes are still being allocated to a player. This is called escalation of committment.
SHC // December 31, 2007 at 3:28 pm
I think that scorers usually get much more credit because they’re more or less the focal points for their teams. And they often initiate the team’s offense, draw double-teams or carry the load, so to speak.
If I was rating players, I wouldn’t strictly follow stats like wins-produced because wins-produced doesn’t explain any conditional variables that would lead to those numbers.
For instance, guys like Ben Wallace, Dennis Rodman live off of other people’s misses and garbage buckets (along with playing tough defense that doesn’t always show up in stats).
These guys may have extremely high wins-produced, but if you had a front court consisting entirely of Ben Wallaces, you will have big problem scoring because he’s got 0 offensive moves. Remember, Detroit was able to win the championship only after they got Rasheed Wallace (whose career WP48 seems to be about average) who can draw double teams in the post.
Using wins-produced to evaluate players, you could also argue that you could substitute Brent Barry for Kobe (who has similar wp48 number to Kobe), and Lakers wouldn’t miss a bit. But the truth is that guys like Barry simply doesn’t have talent to carry scoring load for the team. The guy does a decent job of fitting into the system he plays under, but I bet that his production would be horrible if he was asked to score 25ppg, draw double teams (which he can’t do at all), etc.
I also remember your post about Stephen Jackson being a nearly useless player. But the truth is, Warriors’ records with him and without him are vastly different. It might seem as though Baron Davis and Richard Jason improved last year suddenly after Jackson came to the team, but this is where you actually have to watch games to evaluate how each player’s presence affects the overall team play.
I think that wins-produced does an excellent job of telling you the coefficient for each categorical element in producing wins (i.e. a point is about as important as a rebound, a steal, 1/2 block, etc.) but it completely ignores how a player’s role/presence affects the team’s overall plays.
MC5 // December 31, 2007 at 4:43 pm
Who is Richard Jason?
Kent // December 31, 2007 at 5:02 pm
Dave Berri, Harold Almonte, Owen, TG Randini, Pete, Jasons Chandler and Eshleman, and Jon Posner, Happy New Year and may we have another fun year discussing basketball stats.
Jason // December 31, 2007 at 6:26 pm
Thanks!
Happy new year to all as well.
Mike // December 31, 2007 at 9:27 pm
This guys an idiot for talking about the players being replaced to evaluate the rookies. Obviously it’ll be easier to replace pachulia than ray allen, even a 10 year old can figure that out. And one player doesnt lead the whole team to more wins, its overall talent on the team fool.
Owen // December 31, 2007 at 10:20 pm
Happy New Year!
dberri // December 31, 2007 at 10:27 pm
HAPPY NEW YEAR EVERYONE!!!
Believe it or not, I will be posting a new column in a couple of hours. This is what New Years Eve looks like when you have two young daughters (not that it was much different before I had two young daughters).
Harold Almonte // January 1, 2008 at 3:06 pm
Thanks everybody. Happy New Year. And yeah, we’ll have a lot of fun, we’ll also exercise the mind a lot better than solving crosswords, and we’ll try to give our 2 cents to our beloved game.
Anon // January 2, 2008 at 3:32 am
For comparison, how productive was Lebron James in his rookie year?
Westy // January 2, 2008 at 10:26 am
Good thoughts, SHC.
Meet Carl Landry « The Wages of Wins Journal // February 24, 2008 at 11:51 pm
[...] Durant has a WP48 [Wins Produced per 48 minutes] in the negative range. And this level of productivity has been observed the entire season. If he is elected RoY, that can mean that [...]
Greg Oden or Kevin Durant? « The Wages of Wins Journal // April 2, 2009 at 7:22 pm
[...] December 31, 2007: Should the Rookie of the Year Help His Team Win More Games? [...]