In a recent article for Huffington Post, numberFire.com CEO Nik Bonaddio stated: “RG3: It’s Over”. Bonaddio is asserting that it is unlikely that Washington Redskins quarterback Robert Griffin III (RG3) will ever be able to return to the form that enabled him to win the 2012 Rookie of the Year Award and made him one of the best quarterbacks in the NFL that year. More specifically, Bonaddio claims, “The numbers are quite clear: no quarterback who suffered that bad of a precipitous fall in performance ever recovered.”
While Bonaddio may end up being proven correct, there are several problems with his analysis. More specifically, the numbers are not clear at all. Bonaddio starts with using his company’s Net Expected Points (NEP) metric that examines how many points a player’s team should score given his performance. He then states that only six QBs have ever had a drop in their NEP similar to RG3’s drop after his 2012 season. They are: Steve Beuerlein in 1999, Elvis Grbac in 2000, Jay Fiedler in 2001, Tommy Maddox in 2002, Derek Anderson in 2007, and David Garrard in 2007.
As far can be discerned from the information in the article, these are the only quarterbacks used by Bonaddio in his analysis. Having a sample size of six is a very small number to use in such an analysis. Furthermore, even this small sample has major problems when making a comparison to RG3. They are:
- The average age for when each of the quarterbacks in the sample had their best years is 29.8 years old while RG3 was only 22 years old during his rookie year. Only Anderson, at age 24, was close to the age of RG3 in 2012 during his best season.
- Each of the quarterbacks had played in the NFL for at least one year before having their best season. Only Anderson had played in one season before his best year. The rest of the quarterbacks had played in multiple years before having their best years.
- None of the quarterbacks had a significant injury that could account for the subsequent decline in their performance.
- Each of the other quarterbacks had little mobility. RG3’s success, as stated by Bonaddio, is predicated on his ability to run the football.
- None of these quarterbacks was a Heisman Trophy winner or had the same level of success as RG3 did in college.
At the end of the article, Bonaddio claims “Regardless of what the cause was [of the decline], the effect is obvious and it’s rather tragic.” The cause of RG3’s decline is extremely important, especially compared with these other six quarterbacks. For example, it is impossible to rule out that these six quarterbacks were never good, or at least as good as RG3. They had one good season in the midst of having many relatively mediocre or poor seasons.
In addition, an NFL quarterback’s prime is age 29 with his prime range being 26-30 according to Football Perspective. Again, the average age of the six quarterbacks is 29.8. Given his college and 2012 performances, RG3 could just be a more talented quarterback than any of these other players. Since he has not reached the prime age range in his career, RG3 could also continue to improve as he gets older and gains more experience.
There is also a clear reason for RG3’s decline: his injury history. Bonaddio does point out that RG3 did suffer significant injuries both during the 2012 and 2014 seasons that caused his NEP to decline. However, he does not fully account for the fact that RG3’s poor performances in 2013 and 2014 could be due to injuries and recovery from injuries rather than a decline in his skillset. It is definitely possible that RG3 may never recover his running ability from 2012 or that he is more injury prone than other NFL quarterbacks. However, it is impossible to know what RG3’s best performances can be until that can be proved to be the case or he has actually had played in more games where he has fully recovered from these injuries.
Bonaddio’s analysis does show the potential problem with working with advanced analytics in sports. The NEP could be a valuable metric that provides great insights about the true performance of quarterbacks. However, making assertions beyond the stated use of the metric that rely on small sample sizes with clear confounding variables can lead to problematic conclusions. It may or may not be over for RG3, but it is impossible to tell using the evidence presented in Bonaddio’s article.
Featured image credit: Robert Griffin III on a read-option run during the Redskins 24-16 loss to the Eagles in the 2013 season. Photo by Mr.schultz. CC BY-SA 3.0 via Wikimedia Commons.