It wouldn't be ridiculous to say that baseball has totally embraced sabermetrics. Every team in the MLB employs a team of statisticians that look at each game, each player, even each individual pitch from a statistical standpoint. Basketball is currently in the middle of its statistical revolution. in 2006 MIT started the Sloan Sports Analytics Conference which is co-chaired by the general manager of the Houston Rockets, Daryl Morey. Morey has a done a great deal to popularize advanced statistics in the NBA. This year the NBA.com website started publishing their own advanced stats on their website in addition to the classic (points, rebounds, assists) stats. So if the MLB has integrated sabermetrics into their ballclubs and the NBA is in the middle of adopting an empirical approach to sports, where is the NFL in all this?
The above graph highlights exactly how many times the MLB, NBA & NFL were searched for on Google over the past 10 years. The MLB is shown to be the least popular (at least by Google hits) and the NBA is slightly more popular than the MLB. The NFL on the other hand is almost two times as popular as the other two sports combined.
This graph on the other hand highlights how often people searched for sabermetrics/advanced stats in the three sports. The spike in 2009 is likely attributed to the theatrical release of Moneyball. The NFL in this case is still more popular than the NBA & MLB, however its a much closer. The smaller discrepancy is telling of the lack of advanced stats in the NFL. If the NFL is 3 times more popular than the NBA, wouldn't we expect that similar searches concerning the NFL and NBA maintain that discrepancy? As shown above, the NFL is behind when it comes to sabermetrics/advanced stats, which begs the question - why?
The problem in using sabermetrics in the NFL is rooted in the very nature of the sport itself. In order to understand the impact of a statistic its important to isolate whatever factor you're looking at.
For example, if you wanted to figure out why your bike wasn't riding well you'd look at each part individually. First you'd check whether or not the tires or full, then if your gears were working properly, and so on and so forth. In order to test the impact of a given player in a sport you need to isolate as many of the other factors as possible. This is pretty easy in baseball when you're generally looking at a given pitch which, for the most part, only involves two players - the batter and the pitcher. In basketball it gets a little more complex where now a given often involves at least 6 of the players on the court if not all 10. This distinction between baseball and basketball likely explains why sabermetrics seems to succeed more in baseball than in basketball. More assumptions need to be made with basketball and consequently the data becomes a less accurate model of real life. In the Studying Sabermetrics section we look at some of these assumptions and what those mean when evaluating players.
Okay, so baseball statistics work because there's roughly 2 people involved in a given statistic & basketball statistics work somewhat well because there's 6-10 people influencing a given statistic. So in the case of football a given play involves 22 different players so here we see a lot of statistics fail. With 22 extraneous factors to account for when involving a player how do we evaluate an individual? It's especially tough when so much of what happens on the field doesn't show up on the stat sheet.
The challenge that lies ahead for anyone interested in sabermetrics is how do we adopt the multitude of raw individual statistics (yards, touchdowns, etc.) and convert them into a few meaningful statistics.