Stat Analysis

Advanced analytics on player and team performance

2017 Quarterbacks and Salary

Guest column by Kevin Kolbe

Welcome back to Quarterbacks and Salary, where every year I try to get a little bit closer to figuring out just how much the various quarterbacks around the league are worth relative to each other. We are now on Year 4. The goal of this yearly series is to develop a formula to accurately gauge a quarterback's worth. This will be done by comparing yearly cap hits to statistical performance, using TANY/A (Total Adjusted Net Yards/Attempt) and then by adjusting that base amount according to each player's importance to his offense as measured by plays per game (PPG). The final result is deemed a player's PAYD (Pay by Average Yards and Demand). I first wrote about the subject in July of 2015; you can read that piece for a detailed explanation of my methodology.

I made some pretty significant changes to the methodology this year, but before I describe the actual changes that were made, I want to talk about Peyton Manning's 2013 season. In his second season in Denver, Manning had 709 plays resulting in 5,326 yards, 56 touchdowns, and only 10 interceptions. It was a wonderful, record-breaking season, and PAYD agreed, arriving at a value of \$27.1 million. At the time, I was delighted. It was the highest PAYD value any quarterback had received, even though real-life cap hits have increased significantly since then. It was a truly impressive amount—but had Peyton been just a bit more efficient, that value would have plummeted.

The way PAYD was initially designed was to award the best quarterback (in terms of per-play efficiency) the highest cap hit in that season and use that passer's TANY/A and PPG as the standard by which all other quarterbacks that season are evaluated. This meant that the only way to have PAYD exceed a real-life cap hit was to be the top quarterback and be on the first four years of NFL play, or be slightly less efficient than the leader and have a lot more plays per game. In 2013, Peyton Manning, despite a record-setting year, was only the second-most efficient qualifying quarterback. The top quarterback was Nick Foles, with a TANY/A of 8.36. (Mildly interesting fact: in 2015, Peyton Manning would go on to be the second-least efficient qualifying quarterback, with Nick Foles at the bottom, flipping their 2013 results.)

Even separate from that year, there was too much variation in year-to-year PAYD, as the efficiency of the top quarterback, rather than the performance of the league as a whole, set the bar. In 2016, Dak Prescott had an amazing season with a TANY/A of 8.11, third-best that seeason. The previous year, it would have been first by more than half a yard per play. In addition, the top cap hit was all over the place, and nearly became even more volatile. If Peyton Manning's hypothetically improved 2013 was a sign that I needed a new standard for TANY/A, then Joe Flacco's hypothetical 2016 and 2017 were signs that taking the top cap hit wouldn't hold up. Until a restructure, Flacco was slated for cap hits of \$28.55 million and \$31.15 million over the past two seasons, exceeding the next-highest values by \$5.5 million per year, which would have heavily inflated PAYD for high-performing quarterbacks those years.

To address those issues, I made two changes. First, I set a TANY/A mark I could use to compare all quarterbacks, rather than have one quarterback each year set the standard. To determine this mark, I took the 90th percentile of all quarterback-seasons from 2012 to 2016, which was 7.26. I then adjusted that number in each season based on league passing efficiency. This allowed a compromise for comparing quarterbacks playing in different seasons, while also heavily mitigating volatility from year to year. This lowered the year-to-year change in TANY/A standard from 1.01 adjusted net yards per attempt to 0.39. However, it also lowered the bar every quarterback was measured against. In addition, I compared each quarterback to the average number of plays per game for that season's qualifying passers rather than the top quarterback.

The second change I made was to use the third-highest cap hit from each season rather than the top one. This change, thus far, has resulted in values a bit more comparable in terms of cap changes. Under the old system, the top hit in 2013 was \$20.9 million. In 2014, it went down to \$20.4 million, despite the salary cap being \$10 million higher. I can't say I am thrilled with the new method, as it still doesn't scale exactly with the cap. I considered using a percentage of the cap to set a base value, but that would have left me high and dry once I went back to calculate seasons with no salary cap. It's a shame, because transitioning away from raw dollar figures and into cap percentage had been one of my goals for this year.

The results were, on average, very similar. From 2012 through 2016, the average quarterback's value dropped by \$0.12 million per season in the new methodology, though 2015 and 2016 saw significant adjustments. In 2015, the leading quarterback had been Carson Palmer with a TANY/A of 7.59, but he had a below-average PPG, making him too low a bar to which other quarterbacks could be compared. The average quarterback that year lost \$3.67 million by moving to the new system. Considering that the old method had produced four quarterbacks with a PAYD higher than the highest cap hit (all of whom were outside the top 10 percent of efficiency), I think that change was a positive one in terms of accuracy. In 2016, the average quarterback gained \$2.79 million with the new method, since they were no longer being compared to the highest efficiency mark by a considerable margin.

The two biggest single-season changes both involved Matt Ryan. His 2015 value fell by \$6.2 million in the new system, while his 2016 value jumped back up by \$5.6 million.

With that out of the way, let's look at the most recent season. Alex Smith led the pack in 2017 with a TANY/A of 7.43 (1.12 lower than the previous season's leader, Matt Ryan). He didn't, however, finish with the highest PAYD. Carson Wentz's rookie adjustment brought his PAYD to \$27.7 million, \$0.4 million above that of Smith's. The lowest PAYD value went to Brett Hundley with \$3.6 million, which is the highest mark that the lowest player has earned. This is due to a combination of the league having a poor year for efficiency and all of the least efficient players benefitting from a rookie adjustment.

 Quarterbacks and PAYD, 2017 Player Team Games PPG TANY/A Base Value Rookie-AdjustedBase Value PAYD Cap Hit PAYD-Cap A.Smith KC 15.0 40.0 7.43 \$26.1 \$26.1 \$27.3 \$16.9 \$10.4 D.Brees NO 16.0 39.4 7.18 \$24.3 \$24.3 \$24.7 \$19.0 \$5.7 J.Goff* LAR 14.7 36.1 7.11 \$23.9 \$26.2 \$22.1 \$6.3 \$15.8 P.Rivers LAC 15.6 39.2 7.08 \$23.6 \$23.6 \$23.7 \$18.0 \$5.7 T.Brady NE 15.8 40.6 7.06 \$23.5 \$23.5 \$25.3 \$14.0 \$11.3 C.Keenum MIN 14.3 38.0 6.80 \$21.7 \$21.7 \$20.5 \$1.9 \$18.6 C.Wentz* PHI 12.5 42.6 6.72 \$21.2 \$23.5 \$27.7 \$6.1 \$21.6 B.Roethlisberger PIT 14.9 40.9 6.59 \$20.3 \$20.3 \$22.2 \$18.2 \$4.0 M.Ryan ATL 16.0 36.6 6.58 \$20.2 \$20.2 \$17.6 \$23.8 \$(6.2) M.Stafford DET 15.9 40.3 6.46 \$19.4 \$19.4 \$20.6 \$16.5 \$4.1 R.Wilson SEA 16.0 43.2 6.04 \$16.5 \$16.5 \$20.0 \$14.6 \$5.4 J.Winston* TB 11.9 42.7 5.90 \$15.5 \$17.9 \$21.2 \$6.9 \$14.3 B.Bortles* JAX 15.8 38.2 5.89 \$15.5 \$17.8 \$17.0 \$6.6 \$10.4 D.Prescott DAL 15.9 36.4 5.84 \$15.1 \$15.1 \$13.0 \$0.6 \$12.4 K.Cousins WAS 16.0 39.4 5.83 \$15.0 \$15.0 \$15.2 \$23.9 \$(8.7) Player Team Games PPG TANY/A Base Value Rookie-AdjustedBase Value PAYD Cap Hit PAYD-Cap D.Carr* OAK 14.7 38.0 5.62 \$13.6 \$15.9 \$15.0 \$15.7 \$(0.7) M.Mariota* TEN 14.6 37.0 5.58 \$13.3 \$15.6 \$13.9 \$6.6 \$7.3 T.Taylor BUF 14.1 38.3 5.56 \$13.2 \$13.2 \$12.6 \$9.7 \$2.9 J.McCown NYJ 12.3 38.0 5.48 \$12.6 \$12.6 \$12.0 \$6.5 \$5.5 A.Dalton CIN 16.0 35.8 5.29 \$11.3 \$11.3 \$9.4 \$15.7 \$(6.3) C.Newton CAR 15.8 41.9 5.21 \$10.8 \$10.8 \$12.3 \$20.2 \$(7.9) J.Brissett* IND 15.2 38.4 4.96 \$9.0 \$11.4 \$11.0 \$0.6 \$10.4 M.Trubisky* CHI 12.0 33.5 4.69 \$7.1 \$9.5 \$6.8 \$5.3 \$1.5 E.Manning NYG 15.0 40.9 4.68 \$7.1 \$7.1 \$7.7 \$19.7 \$(12.0) J.Cutler MIA 12.3 38.0 4.61 \$6.6 \$6.6 \$6.3 \$10.0 \$(3.7) J.Flacco BAL 15.2 39.5 4.41 \$5.2 \$5.2 \$5.3 \$24.6 \$(19.3) T.Siemian* DEN 9.8 42.1 4.14 \$3.3 \$5.7 \$6.6 \$0.6 \$6.0 B.Hundley* GB 10.6 35.9 3.93 \$1.9 \$4.3 \$3.6 \$0.7 \$2.9 D.Kizer* CLE 13.5 43.8 3.75 \$0.6 \$3.0 \$3.7 \$0.9 \$2.8 * Base value adjusted for players still on their rookie contracts. Minimum eight games played.

And here is the updated career PAYD-Cap differential. In addition to changing the methodology, all 2012 quarterbacks had their numbers change because I had incorrectly inputted Eli Manning's contract value as \$20.9 million (his 2013 value) instead of his actual value of \$9.6 million. While that significantly decreased his "deficit" (and lowered the PAYD of all 2012 quarterbacks), he is still the worst value by PAYD over the last six years, though Joe Flacco is closing the gap.

 Quarterbacks Ranked by Average PAYD, 2012-2017 Player 2012 2013 2014 2015 2016 2017 Average J.Goff* \$22.1 \$22.1 D.Brees \$20.5 \$21.3 \$16.8 \$19.9 \$24.9 \$24.7 \$21.4 T.Brady \$21.5 \$9.7 \$15.4 \$20.1 \$29.3 \$25.3 \$20.2 A.Rodgers \$18.1 \$20.2 \$21.8 \$11.0 \$24.8 \$19.2 D.Prescott* \$22.7 \$13.0 \$17.9 A.Luck* \$15.0 \$14.2 \$22.4 \$19.8 \$17.9 B.Roethlisberger \$14.5 \$10.3 \$19.8 \$18.9 \$18.8 \$22.2 \$17.4 K.Cousins* \$15.2 \$21.5 \$15.2 \$17.3 C.Wentz* \$6.4 \$27.7 \$17.1 P.Manning \$18.9 \$30.1 \$18.1 \$1.1 \$17.1 M.Ryan \$18.4 \$9.7 \$16.5 \$10.1 \$29.8 \$17.6 \$17.0 M.Stafford \$19.2 \$11.5 \$11.7 \$13.3 \$17.7 \$20.6 \$15.7 J.Winston* \$12.4 \$12.2 \$21.2 \$15.3 R.Wilson* \$11.1 \$11.3 \$14.7 \$18.7 \$15.4 \$20.0 \$15.2 P.Rivers \$6.8 \$16.6 \$11.8 \$14.8 \$12.4 \$23.7 \$14.4 M.Mariota* \$11.4 \$17.1 \$13.9 \$14.1 T.Romo \$16.2 \$11.6 \$12.9 \$13.6 D.Carr* \$5.6 \$12.2 \$21.3 \$15.0 \$13.5 C.Keenum \$6.3 \$20.5 \$13.4 N.Foles* \$24.8 \$1.7 \$13.3 Player 2012 2013 2014 2015 2016 2017 Average A.Smith \$9.3 \$9.8 \$9.6 \$10.7 \$12.7 \$27.3 \$13.2 R.Griffin* \$16.1 \$9.6 \$12.9 J.Freeman* \$12.8 \$12.8 C.Newton* \$16.6 \$10.5 \$9.9 \$17.2 \$9.6 \$12.3 \$12.7 A.Dalton* \$11.3 \$13.3 \$9.9 \$16.3 \$14.5 \$9.4 \$12.5 C.Palmer \$13.9 \$6.3 \$18.1 \$11.5 \$12.5 T.Taylor \$11.5 \$13.2 \$12.6 \$12.4 J.Brissett* \$11.0 \$11.0 M.Vick \$10.6 \$10.6 J.Locker* \$10.2 \$10.2 T.Siemian* \$13.3 \$6.6 \$10.0 B.Hoyer \$8.2 \$11.4 \$9.8 S.Bradford* \$9.5 \$6.1 \$13.1 \$9.6 E.Manning \$11.7 \$0.5 \$13.7 \$13.1 \$9.6 \$7.7 \$9.4 R.Tannehill* \$8.6 \$6.8 \$12.7 \$10.3 \$8.3 \$9.3 B.Bortles* \$2.9 \$13.0 \$12.0 \$9.3 K.Orton \$9.0 \$9.0 C.Kaepernick* \$10.1 \$10.0 \$3.3 \$11.1 \$8.6 B.Weeden* \$8.5 \$8.5 J.Cutler \$7.0 \$9.5 \$8.3 \$11.0 \$6.3 \$8.4 Player 2012 2013 2014 2015 2016 2017 Average R.Fitzpatrick \$8.0 \$7.1 \$9.7 \$13.1 \$2.3 \$8.0 J.Flacco \$11.7 \$1.0 \$11.4 \$8.4 \$9.9 \$5.3 \$8.0 B.Gabbert* \$7.0 \$8.6 \$7.8 T.Bridgewater* \$8.7 \$6.2 \$7.5 C.Ponder* \$7.1 \$7.1 M.Schaub \$11.5 \$2.3 \$6.9 N.Sanchez* \$3.3 \$10.3 \$6.8 M.Trubisky* \$6.8 \$6.8 J.McCown \$0.8 \$12.0 \$6.4 M.Cassel \$4.1 \$4.1 G.Smith* \$1.9 \$5.6 \$3.8 D.Kizer* \$3.7 \$3.7 E.Manuel* \$3.7 \$3.7 C.Henne \$3.6 \$3.6 B.Hundley \$3.6 \$3.6 M.Glennon* \$3.3 \$3.3 K.Clemens \$1.2 \$1.2 B.Osweiler* \$0.6 \$0.6 * At least one season on rookie contract.

 Quarterbacks Ranked by Total Difference Between PAYD and Cap Hit, 2012-2017 Player 2012 2013 2014 2015 2016 2017 Total Dif R.Wilson* \$10.6 \$10.6 \$13.9 \$11.6 \$(3.1) \$5.4 \$49.0 A.Luck* \$11.0 \$9.2 \$16.4 \$1.4 \$38.0 T.Brady \$7.7 \$(4.1) \$0.6 \$6.1 \$15.5 \$11.3 \$37.1 D.Carr* \$4.6 \$11.0 \$19.8 \$(0.7) \$34.7 D.Prescott* \$22.2 \$12.4 \$34.6 J.Winston* \$7.8 \$6.4 \$14.3 \$28.5 M.Mariota* \$7.0 \$11.6 \$7.3 \$25.9 A.Dalton* \$10.1 \$11.9 \$0.8 \$6.7 \$1.4 \$(6.3) \$24.6 B.Bortles* \$(0.9) \$8.3 \$6.4 \$10.4 \$24.2 C.Wentz* \$1.5 \$21.6 \$23.1 N.Foles* \$24.1 \$(2.3) \$21.8 R.Tannehill* \$6.3 \$3.9 \$9.2 \$5.4 \$(3.3) \$21.5 C.Keenum \$2.7 \$18.6 \$21.3 A.Rodgers \$9.6 \$8.5 \$4.2 \$(7.3) \$5.5 \$20.5 T.Taylor \$10.6 \$6.3 \$2.9 \$19.8 T.Siemian* \$12.8 \$6.0 \$18.8 R.Griffin* \$12.3 \$4.8 \$17.1 C.Palmer \$9.9 \$2.3 \$10.7 \$(6.9) \$16.0 J.Goff* \$15.8 \$15.8 D.Brees \$3.1 \$3.9 \$(1.6) \$(3.9) \$7.6 \$5.7 \$14.8 Player 2012 2013 2014 2015 2016 2017 Total Dif R.Fitzpatrick \$2.0 \$4.7 \$6.3 \$9.8 \$(9.7) \$13.1 B.Hoyer \$6.9 \$6.2 \$13.1 T.Bridgewater* \$7.5 \$4.6 \$12.1 J.Freeman* \$11.5 \$11.5 B.Gabbert* \$4.3 \$6.8 \$11.1 J.Brissett \$10.4 \$10.4 K.Cousins* \$14.5 \$1.5 \$(8.7) \$7.3 J.Locker* \$7.3 \$7.3 B.Weeden* \$7.0 \$7.0 A.Smith \$1.3 \$5.0 \$(4.9) \$(5.1) \$10.4 \$6.7 G.Smith* \$1.0 \$4.5 \$5.5 C.Newton* \$11.6 \$4.5 \$2.9 \$4.2 \$(9.9) \$(7.9) \$5.4 T.Romo \$4.4 \$(0.2) \$1.1 \$5.3 K.Orton \$5.0 \$5.0 C.Ponder* \$4.8 \$4.8 B.Roethlisberger \$5.6 \$(3.3) \$0.9 \$1.7 \$(5.2) \$4.0 \$3.7 B.Hundley \$2.9 \$2.9 D.Kizer* \$2.8 \$2.8 M.Glennon* \$2.7 \$2.7 E.Manuel* \$2.1 \$2.1 Player 2012 2013 2014 2015 2016 2017 Total Dif M.Trubisky* \$1.5 \$1.5 J.McCown \$(4.0) \$5.5 \$1.5 K.Clemens \$0.6 \$0.6 M.Cassel \$0.4 \$0.4 C.Henne \$(1.1) \$(1.1) M.Sanchez* \$(9.6) \$8.0 \$(1.6) M.Ryan \$8.8 \$0.1 \$(1.0) \$(9.4) \$6.0 \$(6.2) \$(1.7) P.Manning \$1.4 \$12.6 \$0.6 \$(16.4) \$(1.8) C.Kaepernick* \$8.7 \$6.2 \$(12.0) \$(5.7) \$(2.8) M.Vick \$(3.3) \$(3.3) S.Bradford* \$(3.1) \$(6.9) \$6.1 \$(3.9) M.Schaub \$0.7 \$(8.5) \$(7.8) B.Osweiler \$(11.4) \$(11.4) M.Stafford \$1.4 \$(6.3) \$(4.1) \$(4.4) \$(4.8) \$4.1 \$(14.1) P.Rivers \$(8.5) \$2.8 \$(4.9) \$(6.4) \$(4.1) \$5.7 \$(15.4) J.Cutler \$(2.6) \$(0.9) \$(10.2) \$(5.5) \$(3.7) \$(22.9) J.Flacco \$3.7 \$(5.8) \$(3.4) \$(6.2) \$(12.7) \$(19.3) \$(43.7) E.Manning \$2.1 \$(20.4) \$(6.7) \$(1.4) \$(14.6) \$(12.0) \$(53.0)

Peyton Manning had previously held the record for highest PAYD amount … and still does, now at \$30.1 million. That comes out to 24.5 percent of the 2013 cap. For comparison, the second-highest dollar mark of \$29.8 million by Matt Ryan in 2016 only came out to 19.2 percent. Nick Foles also still holds the record for highest single-season relative value, coming in \$24.1 million above his cap hit in 2013; it will be hard to beat a player on a third-round rookie contract who was the most efficient quarterback in the league. Not counting players on rookie deals, Case Keenum exceeded his pay by \$18.6 million in 2017. Eli Manning's negative \$20.4 million in 2013 is still the worst single-season value, but he no longer holds second place as well, as his brother Peyton was worse in 2015.

That's all for this year. As always, feel free to leave questions and suggestions below. I'm generally pretty prompt at responding.

Guest columnist Kevin Kolbe is originally from Houston, Texas, and has been in the Air Force since December of 2001. If you are interested in writing a guest column, something that takes a new angle on the NFL, please email us your idea at info-at-footballoutsiders.com.

16 comments, Last at 11 Jul 2018, 6:37pm

1Re: 2017 Quarterbacks and Salary

by heylarry21 // Jul 09, 2018 - 4:59pm

Kevin Kolb, you are not fooling anyone by adding an "e" at the end of your name.

11Re: 2017 Quarterbacks and Salary

by Mountain Time ---- formerly Ninjalectual // Jul 10, 2018 - 8:35pm

That is suspicious…

13Re: 2017 Quarterbacks and Salary

by jtr // Jul 11, 2018 - 9:30am

He's working on a long con here, this is all setup to an article about how currently unemployed former Eagles/Cardinals backup QB's are criminally underpaid under the current system.

16Re: 2017 Quarterbacks and Salary

by kckolbe // Jul 11, 2018 - 6:37pm

Wow...way to kill the reveal.

2Re: 2017 Quarterbacks and Salary

by Andreas Shepard // Jul 09, 2018 - 5:48pm

Good stuff.

This gets into more of a philosophical question, but in the next iteration of this I would suggest taking another look at how you handle missed games.

Currently, you are giving "incompletes" to players that don't some minimum threshold for games played. This makes sense from the perspective of evaluating true underlying talent - you can't judge the talent if the player doesn't play. However, if you are judging cold, hard value vs. the contract, you can absolutely judge a player who isn't on the field. If you are paying them and they aren't playing, you are getting 0 value for your money, no matter how talented they are.

Andrew Luck is a good illustration. In 2017 he had a \$19.4M cap hit and played zero snaps. From the Colts perspective, that's zero value, and \$19.4M of overpay (amusingly, only \$0.1M more of an overpay than the Ravens gave Flacco for starting all 16 games). That's very different from Peyton Manning in 2017, who also got an "incomplete". Like Luck, Peyton played zero snaps for an NFL team that year, but unlike Luck he didn't cost anyone any cap space. It's a bit misleading to give both of them the same "blank" for PAYD and PAYD-Cap hit

Tactically, I'd suggest removing your minimum thresholds for games played, and sum up PAYD based on games started. So Luck would get \$0 of PAYD for 2017, and 7/16 games worth of PAYD for his injury-shortened 2015. Instead, I'd suggest limiting your sample to any player who was signed to be a starter. Backups should probably be a separate category, since they are an insurance policy and you don't know how much they will be playing when you sign them.

3Re: 2017 Quarterbacks and Salary

by ssereb // Jul 09, 2018 - 9:27pm

The Colts didn't exactly get zero value from paying Andrew Luck \$19.4 million this season. If they had cut him (let's assume for the sake of argument that they could have done so without paying him, though I don't actually know if that's the case), they would not have the right to get any value from him once he's healthy, i.e. hopefully this season. How much do you think the rights to Andrew Luck's next healthy season would have gone for on the open market? Probably not \$19.4 million, but still a pretty substantial amount.

6Re: 2017 Quarterbacks and Salary

by Andreas Shepard // Jul 10, 2018 - 10:51am

That's true, but I think the right place to capture that future value is in the future years when it happens. If Luck overperforms his contract value in 2018, that will show up when Kevin does this article again next year. But to draw the right conclusions about how well Luck's contract worked out for the Colts, we need to

Take an oversimplified example: let's say Luck had signed a 5 year contract for 2017-2021 with a cap hit of \$20M per year. He didn't play in 2017, and for the sake of argument let's say he bounces back and has an average PAYD of \$25M for the next 4 years. So for 2018-2021, he generates \$5M of value above his cap hit.

If you give Luck an incomplete for his 2017, then this comes out to 4 x \$5M = \$20M of surplus value for the Colts. I think that's misleading. In actuality, they paid 5 x \$20M = \$100M of cap hit, and got 4 x \$25M = \$100M of production. That's a breakeven contract. And the way to account for it is to give Luck 0 PAYD for his 2017 (and therefore a \$20M overpay).

The reason I like this method is it helps feed into the decision the team has to make. In this example, the Colts are offering Luck a contract before the 2017 season. If they think he might miss the whole season, then this methodology says that he needs to overperform by at least \$5M a year in the remaining 4 years to make the contract worthwhile.

7Re: 2017 Quarterbacks and Salary

by Bortling for fun and profit // Jul 10, 2018 - 12:02pm

This accounting is easier and more explainable than finding a replacement level QB to fill in the non-starts. Just assume the replacement level is 0 TANY/A, and multiply the PAYD by the % of games played during the year.

I guess it would depend upon whether we think PAYD is forward-thinking or backwards looking. Rare is the circumstance (Luck excepted) that a front office plans on a QB missing games.

15Re: 2017 Quarterbacks and Salary

by Aaron Brooks Good Twin // Jul 11, 2018 - 1:21pm

8Re: 2017 Quarterbacks and Salary

by kckolbe // Jul 10, 2018 - 12:54pm

I agree completely that games missed due to injury are extremely noteworthy and should be accounted for, but that isn't as straightforward as I'd like. For example, a lot of quarterbacks sit out for the final game of the season. Should that count against them? If not, what if they *did* sit out part of the previous week? Should I assume they could have played?

Separately, what about a QB being benched because the season is over and the team wants to give a young player snaps?

Finally, what if a player misses a game due to injury, but never gets to start again that season because the backup plays well and keeps the starting job (Alex Smith, 2012).

As for your final suggestion, it isn't always clear who was signed to be the starter. Good example of that would be 2015, when the Texans signed Ryan Mallett and traded for Brian Hoyer, paying them about the same money, and frequently changed their minds on who the starter was.

12Re: 2017 Quarterbacks and Salary

by Dan // Jul 10, 2018 - 11:56pm

I agree with Andreas.

The cases that you're worried about are a relatively small fraction of games so I wouldn't worry about them too much. And with some of them, the solution that feels philosophically correct matches Andreas's formula. e.g., Alex Smith was in fact not providing any value to the Niners while he was on the bench; maybe it wasn't "his fault" that Kaep played so well but apportioning credit/blame is not the main point.

As a slightly different implementation, you could just use total plays instead of plays per game. (Actually, I'm not sure if this would produce different numbers or is just a different way of describing it.)

On who to include, you could include any QB that meets at least one of three cutoffs: PAYD at least X, cap hit at least Y, or number of plays at least Z.

14Re: 2017 Quarterbacks and Salary

by kckolbe // Jul 11, 2018 - 11:53am

"Alex Smith was in fact not providing any value to the Niners while he was on the bench; maybe it wasn't "his fault" that Kaep played so well but apportioning credit/blame is not the main point."

This is a really good point. QB value can change based on decisions outside their control. In 2013, when Eli Manning was attempting to run a vertical offense with no offensive line, his PAYD suffered tremendously due to that coaching decision.

4Re: 2017 Quarterbacks and Salary

by Jerry // Jul 10, 2018 - 2:47am

"I considered using a percentage of the cap to set a base value, but that would have left me high and dry once I went back to calculate seasons with no salary cap."

Couldn't you use percentage of payroll in pre-cap years?

10Re: 2017 Quarterbacks and Salary

by kckolbe // Jul 10, 2018 - 12:57pm

That's not a bad idea at all. I'll need to look into some things first, but I will give you credit if I adopt that method.

5Re: 2017 Quarterbacks and Salary

by Bortling for fun and profit // Jul 10, 2018 - 9:59am

Seems like some adjustment for playing time is needed. Take Bortles and Winston, who have similar ANY/A and Base Salary. Winston is credited for \$4M more PAYD, due to a higher attempts per game. However, Bortles actually played 4 more games.

This stat would be more useful if you determined a replacement-level QB (Worst starter? Average backup?) and imputed the replacement-level stats for missing games. Then you wouldn't have to worry about playing time issues, and you can compare an injured elite QB (Watson) with a consistently healthy stat compiler (Bortles).

Health / Non-Health is a skill, and should be paid / recognized as such.