Top ten for WPA in a game? No. Not even close, and the reason is because the Bills never really came close to winning. Josh only got to about a cumulative 45%. Making your team 45% more likely to win in a relatively hopeless loss is an extremely impressive performance, but is a long way outside the top ten.
In this context, we must understand that scoring and winning are slightly different things. As far as helping to contribute to his team scoring points, Josh Allen likely just played the best game ever played in a loss, but did not have an exceptional game as far as helping the Bills win, because he never got the ball within one score with a chance to tie the game, due to the folly of players other than himself, mostly the blocked punt.
As far as a QB doing the most to help his team win, but still losing, that answer remains Chad Henne, but that is strictly because he had a chance to do so. Football is a team game. This recent Josh Allen performance shows more than anything that even the best QB performances will not do very much at all to help a team win if not given the proper circumstances.
Fantastic response. I’m just now hearing about this statistic for football and thinking about it. So to maximize your chances at WPA *specifically* you would need to be in a see-saw game and score every time you had the ball??
There are two ways to do it. Either what you suggested, OR come into the game as an extreme underdog (starting WP around 10%), and put your team in very good (80% or better) spots to win. With very few exceptions, this entails being a really bad player (to trick the WP model into thinking you're incapable of coming back), but playing great. To become the highest WPA loser ever, both WPA maximising strategies must be employed at once, like what happened to Chad Henne in this game, but the top ten is still achievable with either strategy.
The real outlier here is actually Matt Schaub, because this IS NOT the way to maximise WPA at all. Coming into the game as a prohibitive favourite, throwing two different back breaking INTs, and coming perilously close to both losing and tying the game is not what I would've envisioned as the highest WPA game of all time. It also doesn't (until the fourth quarter) employ either strategy we just discussed. The Texans were the best team in the NFL, and prohibitive favourites, so don't get any free WPA just for playing well like a bad team would, and this game also wasn't really back and forth. The Jags didn't crack 50% WP until deep into the third quarter.
Nevertheless, a wild fourth quarter and a ludicrous OT were able to overturn all that, so in truth, the real strategy for maximising WPA is to come into the fourth quarter down 14 points or more, and put your team in almost a 100% chance to win by the end, but have your team miss the game winning FG, so you can tack even more WPA on in OT.
For a stat like WPA, it's all circumstance dependent, just like it is in baseball. If your team gets down two scores by no fault of your own and you're Josh Allen, too bad. You're too far behind and you lose. If you get down two scores by no fault of your own and you're Matt Schaub, you get to be the all-time leader for WPA in a game.
It's exactly this divergence that causes me to continually insist that QBs ought to be graded based on their abilities in scoring, not winning.
I was clearly off on several details, but "wild November game against the Texans" had me expecting this to end with Mike Thomas catching a Hail Mary -- obviously should've at least remembered that Garrard was throwing that one! Disappointed to find out the Jaguars lost but I bet we'll make a bunch of great picks in the 2013 Draft!!!
I'm not as familiar with the football WPA metric as for baseball but this is quite a wacky illustration of it; is the trick here that there was such a weird combination of "comeback followed by leverage" for both teams? It's gotta be impossible to pull something like this off without the benefit of having one of the participants be a 16-point favorite to skew the win probability right from the outset. A 96% win probability at 7-0 after 3 minutes is hilarious, but having lived through the 2012 Jaguars (among other years) I get it.
The other theory that comes to mind is that the winning and losing QB single-game WPA records have to be somewhat correlated. First Henne gets to add a win's worth of WPA from a baseline of "96% likely to lose," then Schaub mounts a comeback, then they both get multiple drives in game-ending situations where a single play can add .25. Like you mentioned, WPA is asymmetrical with respect to an individual quarterback (since they can repeatedly bail out their defense), but is there any way to put up a WPA well over 1.00 without the opposing quarterback also balling out?
I laughed too. I couldn't believe the 96% when I saw it. A seven point lead should never produce a WP that high, but like I said, the 2012 Jaguars were worse than even a typical first overall pick level roster. It was probably deserved.
As far as your WP assertions, the high WPA for both the winning and the losing sides are absolutely correlated. Four spots in the top 20 WPA of all time are eaten up by just two games, with Mahomes and Allen in THAT playoff game being the other one, although Josh in that game comes nowhere near Chad's WPA in this game. Additionally, two of the worst WPA performances of all time both also happened in the same game. Most of the time, in the event of either exceptional or awful play, the WP will not change very much. It will just get on either zero or 100 and stay there, restricting people from getting on this particular leaderboard, but in circumstances like these two, exceptional play is not rewarded with an easy win, which is how you get all-time WPA greatness.
As far as your other question, the answer is actually yes. You can put up all-time great WPA numbers without your opponents balling out on offence. I did not expect it to be yes. I definitely thought you were right about this, but upon further research, it is very difficult to the point of almost being impossible to put up an extremely high WPA stat without the other side playing super well, but it is possible. However, just like anything else involving an extremely high/low value for a WPA statistic, it involves an absurd confluence of circumstances.
The second highest WPA in a NFL game by any player since the stat began being tracked in 1999 was accomplished by Mark Sanchez, in a 2010 game against the Cleveland Browns. 178%. His counterpart, Colt McCoy, was in the negative, and the Browns generated negative EPA/Play as an offence. Additionally, there was no prohibitive underdog, as the WP model thought the Browns had about a 35% chance to win to start. You may ask how these things can possibly all be true at once, and I wondered that too, so I looked into it.
The Jets missed two FGs that day, including one 24 yarder, each of which would've put the game away. They nearly messed around and lost the game in regulation, but didn't, which gets us to OT, where the Jets miss a third potential game winning FG, and for everything in the world this game looks to be a tie, as the Browns punt the ball back to New York with 24 seconds left, but a deep ball to Santonio Holmes wins the game 26-20.
That is how you have a WPA masterpiece against a team who is not balling out on the other side, and evidently it's the only way you do it, as this is the only game in the top 20 of the leaderboard that I would not describe as a shootout, so for the most part you're correct. It almost has to be a shootout to get to the top of the leaderboard, but one out of 20 is enough to disprove such a strict thesis.
P.S. Don't feel bad about the Henne snub. Like I said, I'm not sure if he played like an all-time great in this game, or if he played really really badly. No other performance by any player in the history of the NFL leaves it so ambiguous as to whether the player played well or not.
What a great edge case, of course it involves the Jets and Browns! And the guy who missed those three field goals was Nick Folk, who has been the NFL's most accurate field goal kicker in 2023 and 2024?
Thanks for doing the follow-up research, thrilled with the results!
Really? Why? I thought you were a Cowboys guy buddy!
The Cowboys were playing a pretty important game (which also went to OT) at the exact same time as this. Why would you ever be watching a game between the NFL's best and worst teams with CBS's D-team on the call?
As the Jaguars fan in the room, I am extremely happy you did, but why would that ever happen LOL?
Did Josh Allen just break into the top ten?
Top ten for WPA in a game? No. Not even close, and the reason is because the Bills never really came close to winning. Josh only got to about a cumulative 45%. Making your team 45% more likely to win in a relatively hopeless loss is an extremely impressive performance, but is a long way outside the top ten.
In this context, we must understand that scoring and winning are slightly different things. As far as helping to contribute to his team scoring points, Josh Allen likely just played the best game ever played in a loss, but did not have an exceptional game as far as helping the Bills win, because he never got the ball within one score with a chance to tie the game, due to the folly of players other than himself, mostly the blocked punt.
As far as a QB doing the most to help his team win, but still losing, that answer remains Chad Henne, but that is strictly because he had a chance to do so. Football is a team game. This recent Josh Allen performance shows more than anything that even the best QB performances will not do very much at all to help a team win if not given the proper circumstances.
Fantastic response. I’m just now hearing about this statistic for football and thinking about it. So to maximize your chances at WPA *specifically* you would need to be in a see-saw game and score every time you had the ball??
There are two ways to do it. Either what you suggested, OR come into the game as an extreme underdog (starting WP around 10%), and put your team in very good (80% or better) spots to win. With very few exceptions, this entails being a really bad player (to trick the WP model into thinking you're incapable of coming back), but playing great. To become the highest WPA loser ever, both WPA maximising strategies must be employed at once, like what happened to Chad Henne in this game, but the top ten is still achievable with either strategy.
The real outlier here is actually Matt Schaub, because this IS NOT the way to maximise WPA at all. Coming into the game as a prohibitive favourite, throwing two different back breaking INTs, and coming perilously close to both losing and tying the game is not what I would've envisioned as the highest WPA game of all time. It also doesn't (until the fourth quarter) employ either strategy we just discussed. The Texans were the best team in the NFL, and prohibitive favourites, so don't get any free WPA just for playing well like a bad team would, and this game also wasn't really back and forth. The Jags didn't crack 50% WP until deep into the third quarter.
Nevertheless, a wild fourth quarter and a ludicrous OT were able to overturn all that, so in truth, the real strategy for maximising WPA is to come into the fourth quarter down 14 points or more, and put your team in almost a 100% chance to win by the end, but have your team miss the game winning FG, so you can tack even more WPA on in OT.
For a stat like WPA, it's all circumstance dependent, just like it is in baseball. If your team gets down two scores by no fault of your own and you're Josh Allen, too bad. You're too far behind and you lose. If you get down two scores by no fault of your own and you're Matt Schaub, you get to be the all-time leader for WPA in a game.
It's exactly this divergence that causes me to continually insist that QBs ought to be graded based on their abilities in scoring, not winning.
I was clearly off on several details, but "wild November game against the Texans" had me expecting this to end with Mike Thomas catching a Hail Mary -- obviously should've at least remembered that Garrard was throwing that one! Disappointed to find out the Jaguars lost but I bet we'll make a bunch of great picks in the 2013 Draft!!!
I'm not as familiar with the football WPA metric as for baseball but this is quite a wacky illustration of it; is the trick here that there was such a weird combination of "comeback followed by leverage" for both teams? It's gotta be impossible to pull something like this off without the benefit of having one of the participants be a 16-point favorite to skew the win probability right from the outset. A 96% win probability at 7-0 after 3 minutes is hilarious, but having lived through the 2012 Jaguars (among other years) I get it.
The other theory that comes to mind is that the winning and losing QB single-game WPA records have to be somewhat correlated. First Henne gets to add a win's worth of WPA from a baseline of "96% likely to lose," then Schaub mounts a comeback, then they both get multiple drives in game-ending situations where a single play can add .25. Like you mentioned, WPA is asymmetrical with respect to an individual quarterback (since they can repeatedly bail out their defense), but is there any way to put up a WPA well over 1.00 without the opposing quarterback also balling out?
Finally, have to plug my posts on the trade that ended the Blaine Gabbert era in Jacksonville (https://tradestenyearslater.substack.com/p/trades-ten-years-later-blaine-gabbert) and the Matt Schaub era in Houston (https://tradestenyearslater.substack.com/p/trades-ten-years-later-matt-schaub). I regret that the Gabbert post includes the line that Henne "wasn't good by any stretch" during a period of starts that I now know included the best-ever WPA by a quarterback in a losing effort.
I laughed too. I couldn't believe the 96% when I saw it. A seven point lead should never produce a WP that high, but like I said, the 2012 Jaguars were worse than even a typical first overall pick level roster. It was probably deserved.
As far as your WP assertions, the high WPA for both the winning and the losing sides are absolutely correlated. Four spots in the top 20 WPA of all time are eaten up by just two games, with Mahomes and Allen in THAT playoff game being the other one, although Josh in that game comes nowhere near Chad's WPA in this game. Additionally, two of the worst WPA performances of all time both also happened in the same game. Most of the time, in the event of either exceptional or awful play, the WP will not change very much. It will just get on either zero or 100 and stay there, restricting people from getting on this particular leaderboard, but in circumstances like these two, exceptional play is not rewarded with an easy win, which is how you get all-time WPA greatness.
As far as your other question, the answer is actually yes. You can put up all-time great WPA numbers without your opponents balling out on offence. I did not expect it to be yes. I definitely thought you were right about this, but upon further research, it is very difficult to the point of almost being impossible to put up an extremely high WPA stat without the other side playing super well, but it is possible. However, just like anything else involving an extremely high/low value for a WPA statistic, it involves an absurd confluence of circumstances.
The second highest WPA in a NFL game by any player since the stat began being tracked in 1999 was accomplished by Mark Sanchez, in a 2010 game against the Cleveland Browns. 178%. His counterpart, Colt McCoy, was in the negative, and the Browns generated negative EPA/Play as an offence. Additionally, there was no prohibitive underdog, as the WP model thought the Browns had about a 35% chance to win to start. You may ask how these things can possibly all be true at once, and I wondered that too, so I looked into it.
The Jets missed two FGs that day, including one 24 yarder, each of which would've put the game away. They nearly messed around and lost the game in regulation, but didn't, which gets us to OT, where the Jets miss a third potential game winning FG, and for everything in the world this game looks to be a tie, as the Browns punt the ball back to New York with 24 seconds left, but a deep ball to Santonio Holmes wins the game 26-20.
That is how you have a WPA masterpiece against a team who is not balling out on the other side, and evidently it's the only way you do it, as this is the only game in the top 20 of the leaderboard that I would not describe as a shootout, so for the most part you're correct. It almost has to be a shootout to get to the top of the leaderboard, but one out of 20 is enough to disprove such a strict thesis.
P.S. Don't feel bad about the Henne snub. Like I said, I'm not sure if he played like an all-time great in this game, or if he played really really badly. No other performance by any player in the history of the NFL leaves it so ambiguous as to whether the player played well or not.
What a great edge case, of course it involves the Jets and Browns! And the guy who missed those three field goals was Nick Folk, who has been the NFL's most accurate field goal kicker in 2023 and 2024?
Thanks for doing the follow-up research, thrilled with the results!
I watched it! :)
Really? Why? I thought you were a Cowboys guy buddy!
The Cowboys were playing a pretty important game (which also went to OT) at the exact same time as this. Why would you ever be watching a game between the NFL's best and worst teams with CBS's D-team on the call?
As the Jaguars fan in the room, I am extremely happy you did, but why would that ever happen LOL?