About two weeks ago, Bill Barnwell published an article on ESPN called, “The NFL stats that matter most.” It was a classic late offseason post. I’m sure he sat down with his editor and said, “I’ll write about my processes and some of the statistics I use to decipher NFL players’ and teams’ performance each week. Since an actual game hasn’t been played in over five months, I don’t have any original material and, this way, I don’t have to do any research!”
Regardless of my cynical projection of the motives behind the article, I found it insightful and helpful in understanding some key statistics for analyzing performance beyond what is written in the box score. One of the most discussed and misunderstood stats is QBR (quarterback rating). It’s discussed because it assesses quarterback performance regardless of scheme or style of play and misunderstood due to the lack of transparency from ESPN on how it’s calculated and how it differs from passer rating, an older statistic. I know what a good QBR is (QBR is measured on a 100 point scale with 50 as an average QBR and 75 representing the QBR of a Pro Bowl caliber QB) but I can’t point to the exact inputs (e.g., number of touchdowns, yard, or interceptions) that separates a bad QBR from a good QBR.
As a sports fan who is not affiliated with any NFL team or involved in the player evaluation process, I posed a question to myself. How does QBR, or any passing statistic, serve as a signal for drafting fantasy football quarterbacks?
It’s easy to say that Tom Brady or Aaron Rodgers are great quarterbacks, but does that necessarily mean they’re the best fantasy quarterbacks? Wouldn’t bad quarterbacks who play in pass heavy offenses still be great options, given the amount of yards and touchdowns they record in garbage time (shout out to Blake Bortles!)?
Similarly, quarterbacks in conservative offenses may not score a lot of fantasy points but are well liked by rate-based statistics. For example, Sam Bradford had the 6th highest passer rating last season (99.3) but averaged the 23rd highest fantasy points per game (14.9) among quarterbacks last season. Bradford didn’t score touchdowns (3.6 % TD percentage) and relied on short passes (his 9.8 yards per completion placed him ahead of only Brock Osweiler) but was efficient and managed the ball well (only five interceptions).
To answer these questions, I compiled the passing statistics for all qualified quarterbacks from the last five NFL seasons. Using this data, I calculated the correlation coefficient for fantasy points per game played with a list of relevant passing statistics, as shown in the table below.
Passer Rating and QBR
The correlation coefficient between QBR, the QB stat that matters most to Barnwell, and fantasy points per game is 0.78, which signals a positive relationship between the two variables. Based on this, it would make sense to draft a QB who you expect to post a high QBR. Similarly, passer rating also has a high correlation, 0.81, with fantasy points per game, suggesting that you could use either QBR or passer rating as a screen.
However, determining how to apply these screens is tricky. QBR and passer rating judge quarterbacks differently. Conservative quarterbacks (or quarterbacks playing in risk averse offenses) tend to have higher passer ratings and lower QBRs, as passer rating does not consider the game situation in which each throw takes place. Let’s consider Sam Bradford’s 2016 season again, for example, as he ranked 6th in passer rating but 17th in QBR. While Bradford led the NFL in completion percentage by completing 71.6% of his passes, he was also second to last in yards per completion. Passer rating doesn’t care that his passes didn’t go anywhere. It just sees an efficient quarterback who completes a lot of passes, even though he gained 6 yards on third and 7.
QBR, on the other hand, rewards quarterbacks for creating scoring chances and scoring points efficiently within the context of the game being played. ESPN does this through its Expected Points Added metric, which measures how much each completion contributes to the amount of points a team is expected to score. Therefore, a 7 yard on pass on third and 6 could boost QBR more than an 8 yard pass on third and 9, depending on other factors.
In addition, QBR also does not treat all interceptions equally. Bradford threw only five interceptions, which boosts his passer rating and QBR. However, other quarterbacks may not be penalized as much for their large number of interceptions, if it occurred in meaningless situations (as measured by win probability at the time the pass was thrown) or was caused by a drop by a wide receiver.
Finally, QBR also accounts for expected points added from scrambling. Tyrod Taylor, a so-so passer who relies on his rushing abilities in certain cases, ranked 18th in passer rating but 9th in QBR and fantasy points per game in 2016. Passer rating overlooks rushing ability, which provides a supplemental fantasy football benefit.
With respect to evaluating quarterbacks, QBR built upon passer rating and adjusted it to account for the context of the game. For fantasy football purposes, it makes sense to target quarterbacks who rate highly in both measures, as this usually signals a good quarterback. But if you’re not willing to pay a premium for a quarterback and are looking for steals in later rounds, consider a quarterback like Tyrod Taylor, who makes plays using his legs, which boosts his QBR and total fantasy points, but not his passer rating. In contrast, since passer rating doesn’t care about game situation, it also makes sense to consider a quarterback with a high passer rating but a low QBR if the QB scores a lot of touchdowns in garbage time.
 My personal opinion is that statistics like these should not be used as the sole input for deciding who to draft or sign on the waiver wire. They should be used as a screen (remove all the bad ones and make sure to draft a QB before all the QBs who pass the screen are gone) or as one input among other qualitative factors.