This is going to be an evolving post. We are doing a series of posts and podcast episodes over the summer that focus on the upcoming NFL season. The foundation for these episodes is a new QB performance metric called QB Wins. As the series progresses I will update this post to flesh out the details and logic for this metric.
UPDATE: Check out how Professor Lewis leveraged the QB Wins metric to project 2022 season outcomes for all 32 NFL teams.
Post 1: QB Wins and some Player Rankings
In broad terms, the challenge in creating an advanced statistic is choosing the information to use, deciding how to weight the information, and linking the data to something that objectively matters. I’m not going to go into the algorithm specifics, but my approach uses multiple types of QB performance data (yards, TDs, interceptions, etc…), other team-level data (running games, and defensive stats), and expert-derived statistics (QBR, NFL Passer statistic).
Including the broad diversity of data allows the individual data elements to be reweighted and for the inclusion of “expert insights”. Bringing in defensive and running data allows these factors to subtly influence QB performance. The data is combined using a statistical technique that reweights the data and provides controls for things like defensive performance. Another set of statistical models links QB performance to winning while controlling for other factors. The approach is evolving.
The QB Win metric is best illustrated through implementation. Table 1 shows the top ten QB Win scores for the last 3 years. Over the previous three years, the top QB performance was Drew Brees with a 4.66 QB Win score. This score suggests that Drew Bree was worth 4.6 games more than an average QB in 2019. The Saints were 13-3 in 2019. Replace Brees with an average QB, and the Saints are an 8-8 or 9-7 team. Brees 2019 is followed by Aaron Rodgers 2020 and Patrick Mahomes 2020. The table also tells us that the true elite QBs are worth about 3 incremental games.
Table 2 shows the worst-performing QB seasons over the last three years - Justin Fields, Zach Wilson, Carson Wentz, and two years of Sam Darnold. The rookie class of 2021 had a particularly hard transition and produced the two lowest scores. Fields and Wilson (like Brees and Rodgers) are outliers. The best QBs are worth about 3 wins while the worst QBs score at around -3 games. The worst scoring QB in 2019 was Baker Mayfield at -2.5 games.
There aren’t a lot of surprises on these lists. All-time greats at the top and rookies and busts at the bottom. The estimated win effects are interesting. Elite talent is worth about three games, while a disaster at the QB position is about minus four games. Suppose we had a completely average NFL franchise, an 8 and 8 type team (8.5 and 8.5 now). Having Aaron Rodgers come to town turns them into an 11- or 12-win team. Having Sam Darnold as the starter makes them a 4- or 5-win team. Tampa Bay was 7 and 9 in 2019. Tom Brady arrived in 2020, and they went 11-5.
Also streaming on Apple Podcasts, Spotify, and Stitcher.