by Cameron Dowd
The wave of the young signal callers is upon us. This year there are five rookies and five second-year quarterbacks starting from opening day. Over the past ten seasons there has never been a year in which this many first and second-year quarterbacks started the majority of games for their respective teams. The high number of inexperienced starting quarterbacks got us thinking: is the league in a flux with so many teams looking to rebuild, or does experience not matter to quarterback play?
To answer this question, we wanted to determine how experience affects quarterback performance. If season of experience is not a significant predictor of quarterback play, then perhaps even teams in contention for a playoff spot are justified in starting an inexperienced quarterback.
To measure performance, I collected data on expected points added per play (EPA/P) from the past ten seasons for quarterbacks who started a majority of their team’s games. In addition to EPA/P, I recorded the players’ years of experience, age, and if and where the player was drafted as well.
After collecting the data, I ran a regression using a 95% confidence level to determine if experience, quarterback age, and when and if the player was drafted was a significant predictor of quarterback performance as measured by EPA/P.
Here are the results:
Experience, age, and if the player was drafted all had a significant effect on player performance. Controlling for the included variables, an additional year of experience increased EPA/P by about 0.02. The inclusion of age is especially important, as we are focusing on the effect of experience on production. Having age in our regression allows us to compare how much better we would expect a 7-year veteran and 28-year-old quarterback, like Aaron Rodgers or Alex Smith, compared to a rookie 28-year-old quarterback, like Brandon Weeden.
Surprisingly when the player was drafted did not have a significant effect on performance. There are two potential explanations as to why draft position did not have a significant impact on player performance. One reason is the high bust rate at the quarterback position. Players such as JaMarcus Russell and Tim Couch, both taken with the first overall pick in their respective drafts, are classic examples of players failing to perform to the level of their draft position. The second reason is the play of quarterbacks such as Tom Brady and Marc Bulger, who consistently performed at a top level despite being taken with 6th round picks.
As the data shows, experience does have a significant effect on quarterback performance, suggesting the league is indeed in a flux with a high number of teams looking to rebuild. These teams are inserting inexperienced quarterbacks even though more experienced quarterbacks perform significantly better. Perhaps we will see these young signal callers take the league by storm and thrive despite their lack of experience; however based on the past ten seasons though such expectations should be tempered. Almost a third of league is starting quarterbacks with a year or less of experience while these teams may be rebuilding now, we might see their rise to prominence in a few years behind their more experienced quarterbacks.
Great post.
The one thing I might suggest is some kind of era/season adjustment. Over the last few years passers have had higher and higher levels of production. On the other hand, that creates a chicken-egg problem: Are rookie QBs doing better because passing is easier, or does passing appear easier because we have so many talented QBs currently in the league?
PS The explanation for why draft order does not correlate with performance is due to selection/survival effects in the population.
One other thing while it’s on my mind. Age is important but it’s not uniformly linear. In the early years it will have a positive slope and in the later years of a career it will have a negative slope. But again, there will be a survival effect where only the very best QBs ‘survive’ to play into their mid/late 30s, masking the true rate of decline due to age.
Doesn’t looking at players who started the majority of season-team games lead to a survivorship bias where you’d expect experience to have a positive effect because the worst inexperienced QBs would be replaced before being bad experienced QBs?
The reason why I only included quarterbacks who started the majority of their team’s games was to eliminate small sample size bias. For example I didn’t want to include Tom Brady’s 2008 season where he only played 1 quarter.