With the 2016 regular season in the books, I thought it might be interesting to look at some of our homegrown quarterback stats. Normally, I would present Total Adjusted Yards per Play (TAY/P) and New Total Adjusted Yards per Play (NTAY/P)1 and leave it at that. However, we have made those stats available all year and wouldn’t be giving you anything new by going the traditional route. Instead, I’d like to present those metrics with adjustments to account for the defenses quarterbacks have faced. This way, we can put both the gaudy and the paltry passing stats in better context.2 I’d also like to use ESPN’s revamped clutch-weighted Expected Points Added per Play in the same manner.
The methodology is pretty simple. We start by translating the raw quarterback stats in each game into individual game TAY/P and NTAY/P. Afterward, we use Pro Football Reference’s Simple Rating System model to adjust those metrics for the defenses the quarterbacks faced. Keep in mind, these adjustments were made to measure how well defenses performed in these metrics, specifically, so we could establish proper baselines or measuring quarterback performance relative to expectation.
Once we found each team’s defensive SRS scores, we normalized them to league average. In other words, if a team posted an SRS score of -1.55 in TAY/P, and the league average is 8.55 TAY/P, the team’s normalized TAY/P allowed is 7.00. That means any quarterback who plays that team will be measure against 7.00 instead of 8.55. So if a quarterback had a 7.50 TAY/P in a game against them, he would earn a -1.05 value per play without adjusting for opponent but a +0.50 value per play after adjustments. This means guys won’t get unfairly penalized for playing the Broncos and Giants, and he won’t get unfairly rewarded for playing the Browns and Jets.
Make sense? Great. Let’s look at the results.
Team Defensive Stats
This table displays the SRS-adjusted stats for each defense, based on how they fared against opposing quarterbacks only. It is sorted alphabetically, but you can sort by any column. Read it thus: After adjusting for opposing quarterbacks faced, the Arizona Cardinals allowed a true TAY/P of 7.19, a true NTAY/P of 6.69, and a true EPA/P of 0.097. Note that a lower number indicates a stronger defense.
By any quarterback-defending metric you prefer, the Broncos rate the best. The Browns fare the worst by our adjusted yardage stats, but Washington brings up the rear in EPA/P. If you’re not sure what to make of this, think about it this way: with an adjusted TAY/P of 6.36, Denver makes opposing QBs look worse than Brock Osweiler. By contrast, Cleveland makes opposing QBs look like Kirk Cousins.
Use this table as a reference for what follows. Or don’t; live your life the way you want to.
Total Adjusted Yards
This table shows every quarterback with at least one action play in 2016 and is sorted by Total Adjusted Yards over expectation based on defenses faced (VAL). Read it thus: Matt Ryan turned 595 action plays into 7099 Total Adjusted Yards, a rate of 11.93 TAY/P. His defensive strength of schedule was 8.46, meaning he outperformed expectation by 3.47 TAY/P. That gives him a value per-game of 129.0 and a total season value of 2065. If you want to sort by rate stats and filter out non-qualifiers, just type “*” into the table’s search bar.
|Robert Griffin III||CLE||200||1224||6.12||8.41||-2.29||-91.7||-458|
By Total Adjusted Yards over expectation, Matt Ryan laps the field. His 2065 is 67% greater than runner-up Drew Brees‘s 1234, and he has a comfortable margin over Tom Brady in value per game and per play. Ryan’s lead is built on a historically great Y/A and first down percentage, a league-leading touchdown rate, and a solid turnover rate. He took more sacks than you’d like to see, but he made up for it with positive plays.3
Brady, of course, is no slouch. He flaunted his scarlet letter on his way to 11 wins in 12 games, a record breaking touchdown to interception ratio, and his fourteenth division title in fifteen seasons as New England’s primary starter. By TAY/P, he faced the fourth easiest slate of pass defenses, but he almost never played down to the level of his competition.
Dak Prescott and Kirk Cousins both stood out statistically, even if both often appeared to make several good plays at the expense of making great plays. Cousins faced the league’s toughest schedule, by this metric, while the stellar rookie faced the ninth hardest schedule.
Rookie Jared Goff‘s numbers were so bad on a per-play basis that he actually led the league in negative value, despite barely meeting the threshold to qualify for rate stats. Remember, of course, that Goff is really a proxy for Goff learning under a bad offensive coach who was eventually fired, playing behind the Rams line, throwing to Rams receivers, facing the second hardest defensive schedule of any qualifying quarterback.4 Give the guy 20 games before calling his career.
New Total Adjusted Yards
This table shows every quarterback with at least one action play in 2016 and is sorted by New Total Adjusted Yards over expectation based on defenses faced (VAL). Read it thus: Kirk Cousins turned 647 action plays into 5572 New Total Adjusted Yards, a rate of 8.61 NTAY/P. His defensive strength of schedule was 6.80, meaning he outperformed expectation by 1.81 NTAY/P. That gives him a value per-game of 73.3 and a total season value of 1173. If you want to sort by rate stats and filter out non-qualifiers, just type “*” into the table’s search bar.
|Robert Griffin III||CLE||200||998||4.99||6.90||-1.91||-76.6||-383|
This list isn’t going to be too much different from the previous one. The biggest difference is that quarterbacks who rely heavily on YAC from their receivers are going to slip, while those who throw deeper passes are going to rise. Matthew Stafford, who threw short passes, was ninth in TAY/P but falls to eleventh in NTAY/P. Jameis Winston, Tyrod Taylor, and reigning MVP Cam Newton all aired out the ball and fair better here, where the degree of difficulty of their passes is rewarded. Alex Smith and Sam Bradford see drops concomitant with their low percentage of air yards.
Again, Ryan leads the pack by a mile, while Brady drops to fifth in total value. Brady is still second in per-game value, but he drops to third in per-play value because Prescott’s strength of schedule was greater.
My preseason pick for MVP was Russell Wilson. Playing through injury behind a not-ready-for-primetime offensive line, Wilson did not quite reach those heights. His rushing ability, which until this year was among the most valuable of any quarterback’s in history, was gone. His trademark elusive scrambling has often bought him time his line couldn’t provide, enabling him to take deep shots downfield with success, but injuries mitigated that as well.
Expected Points Added
This table shows every quarterback with at least one action play in 2016, according to ESPN’s definition of action plays when calculating Total QBR. It is sorted by Expected Points Added over expectation based on defenses faced. Read it thus: Aaron Rodgers turned 749 plays into 115.4 EPA, a rate of 0.154 EPA/P. His defensive strength of schedule was 0.104, meaning Rodgers outperformed expectation by 0.050 EPA/P. He finished the regular season with 2.33 EPA added value per-game and 37 EPA value for the season. If you want to sort by rate stats and filter out non-qualifiers, just type “*” into the table’s search bar.
|Robert Griffin III||CLE||211||11.2||0.053||0.097||-0.044||-1.87||-9|
The first two tables are based on mostly the same inputs, so they obviously are quite similar. Since TAY/P is based on expected points, there is much overlap with EPA/P as well. However, ESPN has a reverse clutch index to their totals,5 and they also include EPA from penalties. adjusted yardage-based stats do neither of those things, so it is interesting to note the differences here.
The most notable change is the position of Aaron Rodgers and Matthew Stafford in the top three of total EPA. When you consider the number of close games the Packers and (especially) Lions played, it makes sense that a stat that rewards performance in high leverage situations would rate them highly. Of course, the two ran over 700 plays apiece to get there, so their EPA/P only ranks fourth (Rodgers) and fifth (Stafford), behind Prescott, Ryan, and Brady.
Prescott’s lead over Ryan disappears once schedule strength is factored in, but his contributions as a rookie have been remarkable.
Brady faced the second easiest schedule in the league, by this metric, which allows Rodgers to pull up almost even with him. Given that Rodgers had both more plays and more plays per game than did Brady, the Green bay star actually surpasses Brady in both total value and per-game value.
Looking at total value, Ryan led the league by a significant margin, followed by Prescott with Rodgers close behind. However, after accounting for playcalling and offensive support, the race for a certain award may be closer than many in the stats-only crowd are making it out to be. Of course, that’s just my opinion; I could be wrong.
- Total Adjusted Yards per Play with only half credit awarded to quarterbacks for receiving yards gained after the catch. ↩
- Though not perfect context, as that would include factoring in weather, home/away splits, supporting offensive cast, defensive help, coaching, and dumb luck plays. ↩
- Yes, I know, except at the end of that game. Don’t make the mistake of focusing on his worst moment at the expense of all his great moments that made Atlanta a playoff team. ↩
- Similarly, Matt Ryan is a proxy for Ryan playing for a good play designer/caller, playing in a dome, throwing to a good receiving corps, handing off to talented backs, and playing a difficult schedule. Brady is really Brady playing behind a solid line, coached by a Rushmore head coach, playing in a system defenses rarely seem to be able to figure out, playing a light schedule, throwing to decent receivers, playing outside in a northern city. Every player at every position has stats that are really indicative of more than the individual contribution. Good luck figuring out how to parse it. ↩
- By “reverse clutch weighted,” I mean that they don’t give extra weight to high leverage plays (as they used to), but they do reduce the weight of plays made in garbage time. Thus, a quarterback whose team often has a big lead or a deficit will see his contributions mitigated statistically. Alternatively, a quarterback whose team plays in many close games will receive more credit for making the same plays. ↩