It looked like Al Harrington was ready to leave Atlanta for the Indiana Pacers a few weeks ago. But before a deal was signed, Harrington fired his agent. A few days ago his new agent, Arn Tellem, declared that Harrington would consider offers from all teams.In making this announcement, the Associated Press reported that Tellem made the following observation about Harrington’s ability:
“Al is going to be an extremely valuable addition wherever he lands,” Tellem said. “He has already cemented himself as one of the NBA’s up-and-coming young stars.” Tellem said Harrington is much like Utah’s Andrei Kirilenko — someone who can play multiple positions and is effective on both ends of the court.”
It is true that both Harrington and Kirilenko play multiple positions. Each logs minutes at small forward and power forward. But is it true that Harrington offers the same level of productivity as Kirilenko?
To answer this question, we need to measure the productivity of these two players. Let’s begin by just looking at some basic stats. The table HERE reports for a variety of statistics the past five year averages for Harrington and Kirilenko. This time period covers Kirilenko’s entire career and the years where Harrington has been a full-time player.
In looking at these numbers we see some similarities. Across these five years Harrington scored a bit more per game. This past season Harrington’s advantage in scoring even increased. Although Kirilenko averaged 15.3 points per game, Harrington averaged 18.6 points per contest for the Hawks.
On a per-minute basis, Harrington maintains his edge in scoring. And in terms of rebounds and turnovers, both players are essentially the same. So if our player evaluation was restricted to points, rebounds, and turnovers, then Tellem looks to be correct. In fact, Harrington looks to be a bit better.
Unfortunately, players do more than just score points, collect boards, and commit the occasional turnover. To understand productivity you need to consider all the statistics tabulated for the players. And when we move to the other stats, the difference in these players becomes quite clear.
Let’s begin with shooting efficiency. Kirilenko may score a bit less, but he scores fairly efficiently. At least his points-per-shot matches the league average. Harrington, though, has a scoring efficiency well below the league average. And it is not just his five-year average that indicates Harrington scores inefficiently. Harrington has been an inefficient scorer every single year he has played in the NBA.
Beyond shooting efficiency we also see that Kirilenko offers more steals and assists per-minute. And then there are blocked shots, where Kirilenko is one of the league leaders while Harrington is actually below average.
Now what do these differences mean? To answer this question we turn to Win Score, our simplified metric of player performance that neatly summarizes the value of a player. Since Harrington and Kirilenko play essentially the same positions – as Tellem noted, both play small forward and power forward – it is appropriate to compare the Win Score of these two players. Before we turn to the numbers, let’s note that the average small forward posts a Win Score per-minute of 0.152. The average power forward has a per-minute Win Score of 0.215.
Given these benchmarks, how do Harrington and Kirilenko stack up? The answer is reported HERE, where each player’s Win Score for each of the last five seasons is reported. The results indicate that if Harrington is a small forward, he is a bit below average. As a power forward, though, he is well below average. And on the teams that employed Harrington he has logged at least some time at the four spot. So Harrington is below average. And this basic story we would tell if we looked at Harrington each season, or his average performance across all five seasons. In contrast, Kirilenko is above average. And this is the result regardless of which position he plays. Furthermore, Kirilenko has been above average every year he has played in the league. In fact, if we look at the more complex and complete metric, Wins Produced, we see that Kirilenko is typically one of the top fifteen players in the NBA.
Okay, Harrington is below average and Kirilenko is one of the top players in the league. What factors drive this result? I have noticed people arguing that Wins Produced and Win Score are measures driven by rebounding. Although I have noted the outstanding productivity of such non-rebounders as Reggie Miller and John Stockton, the rebounding story persists in some quarters. Well, Harrington and Kirilenko have essentially the same numbers on the boards and those numbers indicate that each player rebounds at an above average rate for a small forward and a below average rate as a power forward. Yet Harrington’s overall productivity is below average for both positions and Kirilenko’s overall numbers indicate he his above average at both spots.
Given these results, it is clear that rebounding is not driving the Win Score story we observe. What else can cause the difference we observed between these two players? An obvious place to look is blocked shots, where there is a very large difference. Although blocked shots are part of the story, less than 30% of the difference in these two player’s Win Score per-minute is driven by shot blocking. No, the big story is shooting efficiency. Both players can score, but Harrington gets his points inefficiently. Kirilenko is not a tremendous scorer, but he is at least average in efficiency. And when we add in his ability to get steals and assists, we see why Kirilenko produces so many wins and Harrington produces relatively few.
One might ask, what’s the point of this story? So Tellem was clearly exaggerating the value of his client. Don’t agents do this all the time? Tellem wasn’t just exaggerating. He was claiming that a below average performer should be compared to one of the best players in the game.
Let me put this statement in some perspective. Jon Kitna was a free agent quarterback this past off-season. Kitna has been around average in some years of his career, and below average in other seasons. What if his agent declared, though, that Kitna is actually the same quarterback as Peyton Manning? How would people react to such a claim? Or think about baseball for a moment. The Detroit Tigers recently sent first baseman Chris Shelton to the minors. What if Shelton’s agent questioned this move and declared that the Tigers had just demoted Albert Pujols, Ryan Howard, or Jason Giambi?
Such comparisons would have surely threatened the credibility of any agent in football or baseball.
The reason people would have scoffed at these comparisons is because we can simply look at the numbers for Kitna and Shelton and see that neither is one of the top players in their sport. The odd part of the story, though, is that it is possible to imagine – especially if you are a fan of Detroit such as myself – that both Kitna and Shelton could become top players in the future. In fact, Kitna has said he thinks he can be a top five quarterback in the NFL under the direction of offensive coordinator Mike Martz. Given Martz’s success with Kurt Warner, this is not a shocking announcement. And it is not as if quarterbacks who have never posted great numbers weren’t suddenly able to do so later in their careers. Rich Gannon is a case that immediately leaps to mind.
The same is true in baseball. Chris Shelton could return to the majors and hit 40 home-runs and post an OPS above 1.000. It doesn’t seem likely at this point, but such dramatic improvements have happened in baseball. Still, because numbers are better understood in baseball and football, for anyone to argue at this point that either Kitna or Shelton are among the top players in their respective games would be considered silly. We might believe it could happen someday, but so far these players are far from the best.
The story in basketball, though, is very different. And the differences start with the consistency story. We can imagine Kitna or Shelton changing. In basketball, though, this is less likely. Relative to the other two sports, basketball players are much more consistent across time. It is highly unlikely at this point in Harrington’s career that he is going to become one of the top players in the NBA. Similarly, barring injury, it is unlikely that Kirilenko will not continue to post above average numbers. In general, above average players in the NBA are always above average. Yes, there are exceptions, but these are infrequent compared to baseball and football.
Another issue in basketball is that the numbers in the sport are not well understood. Scoring tends to be given too much importance. As we report in The Wages of Wins, free agent salary is primarily determined by point totals in the NBA. Shooting efficiency, rebounds, turnovers, etc…. were not statistically significant in our study. Because Harrington and Kirilenko post similar scoring numbers, Tellem can argue that the players are similar.
Now it should be noted, Harrington will probably not command the salary of Kirilenko on the market. There is one factor that also matters. In addition to scoring, blocked shots were the only other statistical factor we found to impact player salary. And Kirilenko has a huge advantage in blocked shots, and hence will likely still get paid a higher salary.
But shot blocking is not the only reason Kirilenko is more productive than Harrington Again, even if Harrington blocked shots like Kirilenko these two players would still have very different Win Scores. No, Kirilenko is a more productive player because he can also get steals and generate assists. In other words, Kirilenko makes positive contributions in a number of different areas.
When we look at just scoring – which drives NBA salary – or just rebounding – which a few people say is all that matters for Win Score – we would conclude that Tellem is essentially right. When we look at the whole picture – which is what Win Score and Wins Produced actually does – then we see that what Tellem said is quite the whopper.
Unfortunately, it is likely one team will listen. And fans of that team will likely be disappointed to see that Harrington will not generate the same number of wins that Utah sees from Andrei Kirilenko.