Who is the most “underrated” player in the NBA? As I noted a few days ago, the answer to this question requires two metrics. The first metric should capture popular perception. The second should approximate reality. Of course, to make such an argument you have to argue that your reality differs from popular perception (so the underrated story requires a bit of an attitude).
Measuring Popular Perception
The discussion of the “overrated” focused on three measures that appear to capture popular perception: NBA Efficiency, Game Score (John Hollinger’s simple measure), and the Player Efficiency Rating (PER or John Hollinger’s more complicated measure).
When we consider how each of these measures is calculated it appears that we would get a different answer from each. For example, compare the formulas for NBA Efficiency and Game Score.
NBA Efficiency = Points + Rebounds + Steals + Assists + Blocked Shots – All Missed Shots – Turnovers
Game Score = Points + 0.4*Made Field Goals – 0.7*Field Goal Attempts – 0.4*Free Throws Missed + 0.7*Offensive Rebounds + 0.3*Defensive Rebounds + Steals + 0.7*Assists + 0.7*Blocked Shots – 0.4* Personal Fouls – Turnovers
These metrics look to be different. But when we look at the population of players from the 2007-08 regular season, we see a 0.99 correlation between a player’s NBA Efficiency and Game Score value.
PER – as the description at Basketball-Reference indicates – is more complicated than both NBA Efficiency and Game Score. But when we compare Game Score per-minute and PER (a per-minute metric), again we see a 0.99 correlation.
In sum, each of these measures is capturing something very similar.
And that something is scoring. As the following posts on each measure indicates, players who score -whether efficiently or not – tend to look good according to each measure.
NBA Efficiency: Do We Overvalue Rebounds? (November 9, 2006).
PER: A Comment on the Player Efficiency Rating (November 17, 2006)
Game Score: Marvin Williams Makes a Hypothetical Deal (December 16, 2007)
Scoring, as The Wages of Wins argues, is the one factor that drives popular perception. Consequently, metrics that are driven by scoring are also going to be good measures of how players are generally perceived.
The Preferred Measure
With the measures of popular perception once again explained, let me take a slight detour before I get to the underrated. Let’s imagine that you wanted a measure of popular perception. Which of these three should you choose?
The answer depends upon how you view complexity. If you wish people to think you are clever, then complexity is considered a benefit. In other words, the simple tends not to impress.
But in empirical research, complexity is a cost (in time and effort). In other words, if all else is equal, a simple approach should always be preferred to a complex approach. Or to put it another way, complexity is only good if the complexity actually gives you something.
Given this argument, NBA Efficiency should be preferred over either Game Score or PER. As outlined above, NBA Efficiency tells essentially the same story and it is the easiest to calculate. My sense, though, is that PER tends to be preferred to Game Score. And Game Score is preferred to NBA Efficiency. In sum, it looks like people like complexity, even if that complexity isn’t giving them anything.
Okay, enough detours. Let’s get to the question this post is supposed to be addressing. Who is the most underrated player in the 2007-08?
The answer to this question will follow the same approach taken in examining the overrated. Again, we need two reference points. Given that this is The Wages of Wins Journal, our measure of reality (or what passes for reality in this forum) will be Wins Produced. The ranking from this metric will be compared to three measures of popular perception: points score per game, NBA Efficiency, and PER.
Let’s start with points scored per game.
Table One reports – via a comparison of points and Wins Produced – the fifteen most underrated players. Topping the list is Marcus Camby. He is followed by Jason Kidd, Andris Biedrins, Jose Calderon, and Tyson Chandler. Each of these players produced a significant quantity of wins, but generally not via scoring.
Next we turn to the Wins Produced-NBA Efficiency story.
Points-per-game has a 0.89 correlation with NBA Efficiency (0.93 correlation with Game Score). Although this is fairly high, we see some differences in the names reported in Table One and Two. Specifically, Camby, Kidd, and Chandler – who were at the top of Table One – do not appear on Table Two. Although these names disappear, Rajon Rondo, Antonio Daniels, Ben Wallace, Jamario Moon, Antonio McDyess, Biedrins, Calderon, and David Lee appear on each list.
The final table looks at PER. Because this is a per-minute measure, we have to compare the PER ranking to the ranking we see from WP48 [Wins Produced per 48 minutes].
Leading this list is Kidd. He is followed by Wallace, Rondo, Camby, and Moon.
Again, we see familiar names. But the name at the top is again different.
So who is the most underrated? If we add together the difference reported in each table the most underrated player in the NBA for 2007-08 is….. Rajon Rondo. And here are the 15 most underrated players if we consider all three perspectives.
1. Rajon Rondo
2. Ben Wallace
3. Jason Kidd
4. Marcus Camby
5. Jamario Moon
6. Antonio Daniels
7. Antonio McDyess
8. Andris Biedrins
9. Samuel Dalembert
10. Tyson Chandler
11. David Lee
12. Jose Calderon
13. Nick Collison
14. Al Horford
15. Josh Childress
A last note on Rondo… when we think of the Celtics we tend to focus on Kevin Garnett, Paul Pierce, and Ray Allen. But in the regular season Rondo produced more wins than Allen. And in the post-season he has offered more than wins than every Celtic not named KG.
This post completes the columns on the overrated, overpaid, and underpaid. For those who missed the previous columns, here are the links to each:
Our research on the NBA was summarized HERE.
Wins Produced, Win Score, and PAWSmin are also discussed in the following posts:
Finally, A Guide to Evaluating Models contains useful hints on how to interpret and evaluate statistical models.