Down 7-0, just 4 minutes into the game, the Secretary of our Trent Dilfer Club, Andy Dalton, stepped onto the field for the first time. After a 1-yard run, 10-yard holding penalty, and incomplete pass, the Bengals were backed into a 3rd-and-19 on their own 14. Dalton drops back, hits Andrew Hawkins who goes for exactly 19, converting to a new set of downs. Here's a look at how the drive unfolded using our Markov model:
Before the 3rd-and-19, there was an 85.3% chance of a punt and 3.1% chance of scoring a TD or field goal. In fact, the second greatest probability of a drive containing a 3rd-and-19 from one's own 24 is that of a fumble at 4.4% (third is an interception at 3.1%). But, Dalton laughed in the face of probability and converted, raising his chances of scoring to 31.9%.
Late in the drive, the Bengals had a 2nd-and-Goal from the 4 - which brings a scoring probability of 91.1%. This time, the Browns defied the odds, stopping two straight run attempts - including 4th-and-Goal from the 1, which ends in a turnover on downs about 15% of the time. Using the 4th down calculator, we know that Marvin Lewis made the right decision. The break even point both in terms of expected points and win probability is right around 30% (compared to an estimated 68% conversion rate).
After the stop, the Browns were backed up against their own end zone. A quick 3-and-out led to great field position for the Bengals, starting at the Cleveland 47. 3 plays later Cedric Benson ran around the left end for a 16-yard TD to tie the game.
In a game that was ultimately decided by 3 points, a 3rd-and-19 conversion that resulted in a +2.57 EPA was one of the forgotten game-changers. Dalton improves his record to 7-4 despite a mediocre +0.02 WPA for the game, fortifying his Trent Dilfer Club credentials.
Keith Goldner is the creator of Drive-By Football, and Chief Analyst at numberFire.com - The leading fantasy sports analytics platform. Follow him on twitter @drivebyfootball or check out numberFire on Facebook.
Nice analysis. I was thinking about this while trying to explain to my father-in-law why Tebow is not the second coming of John Elway. He kept throwing out the "He just knows how to win" line. It occurred to me that maybe in the face of being amazingly mediocre overall, the Dilfer club quarterbacks really are "clutch" players. That is, could we look at the highest EPA/WPA plays in a game and spot plays that could, like this one, be attributed to a particular player performing well above their own average in a key situation?
Wouldn't handing off through the second half of a 27-10 win have been something that better fortified the comparison to Dilfer?
Last week you said:
"Dilfer could not win games for the Ravens, but he could lose them."
Dalton had a big second half, and dominated the second half comparison with McCoy.
Dalton: 12-20-0-177, 0 sacks (14.8 y-p-completion)
McCoy: 8-15-1-49, 1 sack (6.1 y-p-completion)
Down 17-7 at the half, Dalton led a 3-play TD drive that had completions of 8, 35, and 22. Then the game winning field goal was set up by a 51 yard pass.
Isn't that being a quarterback who can win a game rather than being a manager who stays out of the way?
Similar story with Sanchez, the president of the club. He was 9-15-0-114 in the second half vs. Buffalo, with 2 TD passes including a late game-winner (12.6 yards per completion).
I can see Alex Smith (grinder wins, and the 13-6 loss last week matched the Dilfer losses in his Super Bowl year)...and Tebow with the grinder wins...though some would argue Tebow does "win" games as long as you can get him to the last five minutes within one score.
Dalton goes 65 yards in 3 plays in the air for the second half TD that gets his team back in the game...then has a 51-yard pass play that sets up the game winning field goal. Was anyone watching in person thinking, "Yup, just like Trent Dilfer...can't win games but can lose them?"
Jeff-
There will always be outlier cases. Dilfer himself I'm sure had a few great performances (one's you've mentioned), but in general, he fits that archetype.
Appreciate the response. I'll have to respectively disagree. I don't think he fits the "could not win games" trait at all...and he's already established as a rookie that it's not an outlier when it happens. Four of the six wins he finished involved fourth quarter comebacks.
*Versus Buffalo, he led a come-from-behind tying TD drive with 50 yards of passing in an 80-yard fourth quarter drive...plus he ran it in himself from the three-yard line. Then...he led the game winning field goal drive with 32 yards of passing in a 50 yard drive that also saw him scramble for a first down on 3rd and 3 that had to be near the final minute mark (drive started at 1:48).
*Versus Jacksonville, he converted a 4th and 6 on the go-ahead drive in the fourth quarter that gave them the lead for good just before the two-minute warning.
*Versus Tennessee he passed for about 60 yards of a 78-yard drive that gave the Bengals a fourth quarter lead on the road in a game they would eventually win.
Just don't see that being the archetype of "could not win games, but could lose them." Nor do I see why yesterday's game would be seen as an outlier.
Jeez, does this place have edit mode? Respectfully disagree...
The two games where Dalton had a significant impact on winning the game were against Indy and Tennessee (http://wp.advancednflstats.com/playeryear.php?playerid=14-A.Dalton&pos=QB&year=2011), both had WPA over +0.40. Even if he led game-winning comebacks against Jacksonville, Buffalo, and Cleveland, he also put them in a whole to begin with (0.02 WPA, 0.03 WPA, 0.02 WPA). That's not winning the game for your team.
If you put your team in a poor position and then help to get yourself out of the whole but come up even, that's doing your job.
Thanks for saying "out of the whole" so I wouldn't feel bad about "respectively."
Who came up even? Dalton led the game winning drives. "Could not win games" doesn't apply if he's the driving force behind the hole digging and then the rally. He's just making it hard to win games, but he's still winning games.
While you were writing up your response, I was running through the player charts you linked to in the other article, trying to learn Klingon on the fly. Was wondering why you used .50 WPA as the first cutoff rather than .35. Dilfer peaked at .33 in the year you referenced. Why not .35 or .40 to find the best matches? Lifting it up to .50 (about a 50% increase from Dilfer's 2000 peak) could bring in the type of QB you're trying to avoid. Meaning...the high risk/high reward guys who can crack .40 when sailing is relatively smooth but will still have at least 40% of their games under zero.
If you had used .35 to get the best possible matches, Dalton wouldn't have been there...Sanchez would have been there within this year's short sample, but not in his career (he had a 1.03 and an .86 last year, and even .37 and .38 in his rookie year)...then would have fallen out after yesterday with his game vs. Buffalo. Alex Smith and Tim Tebow, the two most clear stylistic matches this year in terms of clearly conservative play and a grinder style of winning would still be there.
In the first piece, you said, "These are the QBs who win games by limiting their self-destruction." Sanchez had interceptions in the wins over Dallas, Jax, SD, and Buffalo. He can win by limiting self-destruction. But, he can also win when he's not limiting his self destruction.
Also wanted to mention that I noticed Dalton has now been positive in 5 of his last 7 games, after three negatives to start his rookie season. He may be trending out of the 40% below zero threshold too if this is a sign of moving up the learning curve.
Anyway, it seems to me that Sanchez and Dalton are in similar molds in terms of attacking downfield in a way that leads to turnovers, but comeback potential when they trail (with the ability to break .35 when things are going well). Their teams DO rely on them to produce in the air.
Smith and Tebow aren't attacking downfield, and therefore aren't digging holes with turnovers. If the team has a lead, they're not going to pass enough to crack .35 because the team doesn't need them to do anything. The teams aren't asking them to produce significantly in the air. These are the best fits for what Dilfer did in 2000. These are the best fits for "can't win a game but can lose one." These are the best fits for "limiting their self-destruction." It's hard to argue that Sanchez and Dilfer do a good job of limiting their self-destruction in my view. I think using a threshold of .50 instead of .35 brought in guys who weren't ideal matches because it allowed a couple of "can bail themselves out with downfield passing" guys into a discussion about conservative play.
Thanks for the conversation. Helped me clarify my thinking on the matter even if we're destined to end up on different pages.
Newest corrections...six lines up from the bottom should be "Sanchez and Dalton" not Sanchez and Dilfer. Also, should clarify that I'm thinking of Tebow struggling to win with his arm. That would seem tough. He can plow his team to the end zone very late in a close game. Which means, the best pure match for Dilfer is Smith.
Jeff, I see where you're going and maybe Smith would have been a better choice for President while Dalton and Sanchez should be honorable mentions, but if you left those two out completely it'd be a very short list. I prefer including them and highlighting what quarterbacks had been under-performing to date.
Also, it's unfair to retroactively judge a to-date article when new data appears.
Appreciate your comments James. Agree that it would be a short list...but, that's because there aren't many truly good comparisons to what Dilfer did in 2000. And, if the goal is to study "under-performing," there are better ways to do that too I believe. Also...tough to say Dalton is "under-performing" given that he's a rookie.
Hope it's not being seen that I'm "judging" last week's article based on new data. The debate started in last week's comments because I believed that the description of Dilfer was being mischaracterized. This week KG said that Dalton's performance was "fortifying" his credentials...yet Dalton's performance involved a stellar 3-play long TD drive on 3-passes that tied the game...and then a 51-yard completion to set up the game winning field goal. That would seem to be the OPPOSITE of what was suggested in his description of Dilfer (can't win games but can only lose them).
Then KG said that was an outlier, and I showed that Dalton had made big plays on other game winning drives this season.
If you have a specific reference of mine that bothered you, please let me know what it was and I'll clarify if, or apologize if need be. Wasn't at all trying to say new data has made the old theory obsolete. Smith and Tebow continued to follow the model in terms at least of not winning with their arm. Dalton and Sanchez were already downfield higher risk/higher reward types before this week.
I think we may kind of be debating how to turn numbers into words again. KG's thresholds suggest that 0.50 and above is "can win a game for you," and .49 or below is "can't win a game for you but can lose one."
And, 0.2, 0.3, 0.2 (following a parenthetical of his in an earlier comment) doesn't count as "winning a game" even if you directly led the TD drive that put the team over the top in the final minutes. So, football people would ask, "Is this a guy...who...when given the ball on his own 25-yard down line down by 4 points with two minutes to go can drive the field for the game winning touchdown?" (or variations thereof). Dalton's doing that...but KG and the math see that as just cancelling out earlier negatives, so "other people" deserve credit for the win.
That may be what's behind some of the disconnect here. The phrase "can't win a game for you, but can lose one" generally refers too guys who CAN'T drive the field for a game winning score when needed...but can stay out of the way when other guys win the game for that team. But, then KG is using the exceedingly high "no games above .50" as a numerical definer for "can win a game for you."
Just went to look up the league numbers. Brady leads the NFL right now at .39 per game...and he's had only two games over .50 this season. So, KG has chosen a very high treshold. He basically defined the Dilfer Club as "the team wins, the QB doesn't grade out well at least 40% of the time, and and the QB doesn't have any "superstar" level performances.
Does .49 or below capture "can't win a game for you?" Something in the 40's is rare and terrific. Nobody's averaging in the 40's through 11 games and only Brady, Rodgers, and Brees are in the 30's in per-game average.
It's possible to "win games" the way football people mean it without being a superstar.
Dilfer peaked at .33 in 2000...something around .35 would seem like a better threshold for building a Dilfer club than something so high it takes a huge performance to get there.
The goal with "similarity score" type analysis is to find the best matches. Using .50 brings in guys capable of bigger games than Dilfer had in 2000 (and KG mentioned in last week's article that Dalton had already popped .40 or better twice...and .40 is 20% better than Dilfer's peak right there).
"Underperforming" wasn't a great word choice, I meant more like "underperforming compared to level of hype and typical performance of a QB with the same record".
I also think this week does support Dalton's inclusion on the list, if only because other QBs are crucified for turnovers even when their production greatly outweighs the downside (namely Eli, Romo, and Rivers). Dalton could only barely make up for his earlier errors yet he gets a pass because the team won. What if those quick scoring drives occurred early and all the negative WPA came later, would you still say the same?
I also didn't realize how high a threshold .50 WPA represented, and a quick look at the Top 10 QBs it seems 0.4 is more realistically attainable.
It looks like Dalton probably doesn't really fit the overall mold for this game, but I certainly would include Sanchez. The only reason I might leave him out is because his team might lose its winning record soon.
My turn to make a mistake - In my last paragraph I didn't mean "mold for this game" but "mold for the Trent Dilfer club, this game notwithstanding."
Maybe this is what's creating the problem for me accepting this grouping.
PER-START-AVERAGES (this year excluding Dilfer)
13-24-1-170 (Dilfer in 2000)
17-27-0.5-192 (Smith)
19-34-1-228 (Sanchez)
20-33-1-228 (Dalton)
Dilfer averaged 24 passes per start, and did a lot of handing off because they weren't asking him to win games. Sanchez and Dalton are seen as weapons by their teams, and are being asked to carry a heavier load. (Note that Dalton only played a partial game in the season opener, he's 21-35-1-242 in the games he started and finished, which is nowhere near Dilfer's numbers).
So, if stuff "comes out in the wash" the same in terms of similar WP's...you're still looking at a different club in terms of what was being asked. And, I'd submit that the Dalton/Sanchez club IS capable of winning games for you...rather than just trying to stay out of the way.
I can see why there's kind of a debate between "digging out of a hole" and "winning the game from the start." But, to me, it's still a relevant win if the QB was playing a huge role in the go-ahead score.
Also, using BB's in-game charts, it looks like...at one point... Cincy was:
10% to win in 2nd half vs. Buffalo
12% to win in 2nd half vs. Tennessee
18% to win in 2nd half vs. vs. Cleveland (G2)
22% to win in 2nd half vs. Jax
Dalton went on the field and executed in a way that brought his team back to victory in all four of those second halves. Winning is still winning even if you're overcoming your own obstacles in the process. That's not consistent with "can't win games for you, can only lose."
Enjoying the conversation, even if it's just the few of us (lol)...
I definitely understand where you are coming from, and they are certainly not mirror images of one another. But your last statement needs to be qualified: Winning is still winning even if you're overcoming your own obstacles in the process. I don't believe it can be considered winning the game for your team - that is, solely on your shoulders - if it is your responsibility that the team was losing in the first place.
That's not to say that the person does not deserve some credit for the win, they do. But winning a game for your team means performing above and beyond expectation to the extent that the team would not have had as likely a shot without you. If a player digs a hole and then climbs out, a team would be just as happy never to have been in the hole to begin with.
I can agree with you, however, that Dalton does not purely win games through damage control, given his ability to fight back. I think moving forward, he may not be the ideal member of the Dilfer Club, but throughout the beginning of the season, he was getting more credit for Bengal wins than he deserved.
Those statistics are dependent on the situation and should be adjusted for era, even though it's only been 10 years. In 2000 the average team passed for 25 fewer yards than in 2011, which means he almost exactly matches Smith's statistics.
Also, the Ravens were 12-4 and won 8 of those games by at least two scores (10+ points), so as you said I imagine they were frequently running out the clock, depressing Dilfer's passing stats.
I haven't seen Dalton play much, but Smith definitely hasn't impressed and I've seen Sanchez play enough to know he's not very good. Smith has the 16th most EPA while Dalton and Sanchez are even lower, and the latter two both have negative WPA. While they've all had some good plays, it's tough to say their teams aren't winning despite them instead of due to them.
Looks like we're all on the same page to a degree now, amidst the differing perceptions. Interesting point about the 10-year difference James...and how that lines up Dilfer with Smith (and lessens the distance between Dilfer and the other two). Appreciate both of you taking time for the back and forth...