Why are player statistics tracked? Since the 19th century baseball has been tracking a host of statistics for its players. The same can be said for basketball and football since the inception of the NFL and the NBA. But why is this necessary?
We know in each contest which team won or lost. But what we want to know is why. The statistics are supposed to help us connect the actions of the players to the outcome on the scoreboard, and in the process, tell us why one team is happy and the other is sad.
Alan Schwarz in The Numbers Game detailed how baseball initially used batting average to evaluate players, and then moved on to such metrics as slugging average, on-base percentage, OPS, linear weights, and the work of Bill James and other Sabermetricians. Each step in the process was designed to better inform us which player’s actions caused the outcomes we observed.
In basketball a similar process is in place, although I suspect it is closer to its infancy. NBA Efficiency is much like batting average, easy to calculate but not entirely an accurate reflection of a player’s impact on wins. Wins Produced, the measure we present in The Wages of Wins, is similar to the linear weights model used in baseball. And although Wins Produced improves upon NBA Efficiency, it is not likely the final word on the subject of NBA player productivity. Well, it might be our final word. But we are sure others will make further progress connecting player statistics to final outcomes in basketball.
And then there is football. Football has problems that frustrate the efforts of those who wish to statistically understand the game. Obviously every position doesn’t have a host of easily understood statistics, though Football Outsiders is helping us in that area.
Even if statistics become available for every player, there remains a problem that I am not sure can be easily overcome. The purpose of statistics is to separate the player from the team. Observers know who won or lost. What we want to know is the contribution each specific player made to that outcome. In baseball we can separate the player from the team. I think we can do this in basketball as well.
But the research we present on football in The Wages of Wins tells us that on the gridiron, separating the individual from the team is difficult. As I noted earlier, very little of what a quarterback does in the current season can be attributed to what the quarterback did the previous campaign. And this is the story we tell when we use our metrics, or the measures posted at Football Outsiders. In essence, quarterbacks are consistently inconsistent.
What does this mean? In a given season we see some quarterbacks are good, other are not so good. But the next season we can look at these same quarterbacks and see a very different ordering. Why does this happen? We have argued that how a quarterback’s teammates performs impact the quarterback’s value. In other words, quarterbacks may make their teammates better, or their teammates may make the quarterback better. The stats, though, can’t tell us which story is correct.
Consequently, the player statistics tracked in the NFL do not serve the purpose statistics serve in basketball and baseball. It is not clear that a rigorous analysis of the player statistics in the NFL would ever allow us to separate the player from the team.
And what does that mean for decision-makers in the NFL? Consider the decision the Miami Dolphins made to acquire Daunte Culpepper. Miami is hoping that Culpepper can finally take the place of Dan Marino. And when we look at Culpepper in 2000, 2003, and 2004 we have evidence that the Dolphins have a great quarterback. But in these seasons Culpepper was throwing to Randy Moss. Without Moss in 2005, Culpepper’s numbers declined. In fact, with Moss in 2001 and 2002 Culpepper was below average. Given all this, which is the true Culpepper? Is Culpepper truly a great quarterback, or merely a product of his teammates? And if it is the latter, does Miami have the talent necessary to elicit above average numbers from Culpepper?
In the end, it does not appear that the stats can answer all these questions.
So what is the value of player statistics in the NFL? Well, they give us something to talk about. But as measures of the productivity of individual players, it is not clear that these numbers are up to such a task. And that is something to remember next time someone quotes a quarterback’s numbers and tells you how great – or not so great – so and so really is.
— DJ
Travis
August 7, 2006
DJ,
Is their any evidence that scoring and/or offensive production is the big indicator on how well a player is played in the NFL? I know that pay varies a great deal amoung positions in the NFL. But among QB’s will an efficient QB make money more (on average) then a QB who throws a lot of TD’s while also throwing a lot of intereptions? I guess what I’m getting at is what is the biggest factor(s) when it comes to forecasting how much money an NFL will be played and are GM’s (unlike in the NBA) good a evaluating talent?
I look forward to reading your NFL stuff as the season progresses. Keep it up!
-Travis W
-Travis W
smilerz
August 8, 2006
I’m not sure that this is true, but I’ve always gotten the impression that a coaches performance played a more signifigant role in football than other sports.
If true, and taking into account a larger role for teamwork there is quite a bit of noise in a football game that would make determining a players skill quite difficult.
Rob Bonter
December 29, 2006
The NFL QB ratings system is also flawed in that it includes BOTH completion percentage and YPP. YPP, alone, would suffice, as it factors in short, medium, and long completions, as well as incompletions and interceptions. In other words, seven yards per pass attempt on 50% completions, is the same as seven yards per pass attempt on 55% completions. One gets more consistency, the other gets more yards per completion. The NFL QB ratings system is misleading and out-dated.
Also, the term “turnover” is a joke. On a regression – linear weights basis, an interception has a value of about 5 1/2 points to the intercepting team, whereas a recovered fumble by the defense is worth just about the same as a FG = three points, making the interception about 83% more valuable. The NFL and every broadcast crew in America seem not to know this, because the distinction between the two is never observed.
pierreboni
September 26, 2007
Wow, I don’t know how active this blog is, but you guys certainly do your homework (including the comment poster). Congrats for speaking smartly about football.