Quarterbacks and fourth quarter comebacks, Part II
Posted by Doug on August 7, 2009
A continuation of Scott Kacsmar's guest post from yesterday (here is the link to Part I)....
Last time we looked at 4th quarter comebacks for Dan Marino (the "new king of the 4th quarter comeback") and John Elway. Now we’re going to look at creating a
clear-cut definition of what a 4th quarter comeback is, what a game-winning drive is, and how other QBs have been tracked, and developing a standard for all teams to follow.
The ideal 4th quarter comeback analysis would be to:
1. Identify the games where a comeback (from a 1--8 pt deficit) was possible: this gives you all successes and failures. Just telling me a QB has 10 comebacks does not mean a whole heck of a lot. But if you tell me he has 10 comebacks in 13 comeback opportunities, I can probably say he’s doing a great job. If he has 10 comebacks in 30 opportunities, he might be someone only as good as Jon Kitna.
2. Identify the situation of the drive: time it started and ended, starting field position, number of timeouts, etc. Not all comebacks/drives are created equal. It’s a lot harder to come back from a 4-8 pt deficit with 30 seconds and no timeouts than it is to start the 4th quarter on the 1-yard line, down by a point.
3. Collect the drive statistics: attempts, completions, yards, length and number of plays, etc. Just your usual QB statistics. Obviously going 8/8 for 80 yards and a TD beats going 1/5 for 8 yards to set up a long FG.
4. Create advanced statistics to better understand performance: average deficit, average yards to go, average time left, points per drive, percentage of 3-and-outs, turnover likelihood, “blown saves”, etc. Along with having the number of successes and failures, this would be the most useful part of comeback
analysis. This is how you can begin to answer who’s really the biggest choker in the league. Unfortunately you need solid play-by-play data here, so the number of seasons you can obtain this type of data is greatly limited.
The problem is that we’re still stuck in stage one after all these years. Due to a semantics argument/hiccup/tie-up, no one is able to agree on a consensus definition of what a 4th quarter comeback or game-winning drive in the 4th quarter/overtime is.
Take a look at this game for Drew Bledsoe. The Patriots trailed 13-10 to start the 4th quarter. They got a FG to tie it, later a FG to win it. 4th quarter comeback for Drew right? Well, a look at the play-by-play shows the final play of the 3rd quarter was a failed 3rd down conversion by the Patriot offense. The first play of the 4th quarter was a 21 yard field goal by Adam Vinatieri on 4th down. That means Bledsoe and the offense never took the field in the 4th quarter with a deficit, therefore no comeback opportunity. It’s a game-winning drive only. I do have a fear that for seasons without play-by-play data on the net that these types of things can happen once in a while where the first play of the 4th quarter is a tying/go-ahead FG.
Let’s stick with another Bledsoe example. This time, it’s a game from 1995.
The game was tied at 14 to start the 4th QT. After Bledsoe leads a FG drive, Boomer Esiason throws a TD for a 21-17 Jets lead. On the ensuing kickoff, Dave Meggett fumbles, but sure-handed teammate Troy Brown is there to scoop up the ball and return it 75 yards for a TD and 24-21 lead. Ty Law intercepts Boomer, Curtis Martin puts it away with a 1-yard TD run. Bledsoe should not get credit for a comeback, as it was purely a special teams play that erased the one deficit they had in the quarter. But, by virtue of his FG drive when it was tied at 14, he is credited with a 4th quarter win.
So it can admittedly get confusing. There is a situation that happens once in a blue moon that confuses
even me. What do you do with a game like this one? (And it’s purely coincidental that I reference a game involving Miami and Denver)
The Dolphins trailed 10-0 to start the 4th quarter. Marino’s successor, Jay Fiedler, already was driving the offense into the redzone when the quarter began. Fiedler throws the 11-yard TD pass to Chris Chambers to make it a 10-7 Denver lead. On the ensuing drive, Elway’s successor, Brian Griese, is intercepted for a TD and a 14-10 Miami lead that they never give up. Denver fumbles the kickoff, which Miami turns into another TD for a 21-10 final.
Does Fiedler get credit for a comeback? He did lead the initial TD drive, which if he did not, Miami may never win the game. It is obviously not a game-winning drive, as the defense scored the winning points. However, Fiedler never took the field with a tie or 1-8 point deficit. Is it a comeback you give him credit for? I lean towards yes, but I’m not as confident with this type of situation as I am the others.
What makes this really interesting is Brian Griese never even got the opportunity to lead a comeback from a 1-8 pt deficit because of the fumbled kickoff. Yet the whole reason they were down in the first place was his pick 6 thrown with the lead. That is a mind-bender of an example, and I do not want to see too many
of those kind.
What can we do to define a 4th quarter comeback once and for all? I’m going to lay out several steps to follow when analyzing a game to see if it’s a comeback/game-winning drive.
1. The game in question was a victory: This is the easiest part of tabulating comebacks. You only need to look at games won (you can look at losses if you’re trying to get the number of opportunities, but that’s more time consuming and especially difficult prior to the play-by-play era). When it comes to ties, I think they are worth a look. If it was prior to 1974 (the season the NFL instituted overtime), they could mean more than post-OT ties. For example, this game with Joe Namath was a pretty impressive 4th quarter performance. Down 24-7, he threw two TD passes and led a tying FG to preserve a 24-24 tie instead of a sure defeat. That is a lot better than the tie we looked at in Elway’s
career. But generally, we’re looking at QBs in the OT-era, and a tie doesn’t cut it in today’s game.
2. Some type of offensive scoring drive put points on the board in the 4th quarter while the team trailed by one possession or were tied: It is ok if the
drive started in the 3rd quarter; as long as it finished in the 4th with the offense still on the field (this eliminates the Bledsoe/Cincinnati situation). It is ok if it’s a FG or TD, as long as the offense was on the field for it. It is ok if it’s in overtime.
3. There can be a difference between a comeback and game-winning (GW) drive: FOR IT TO BE A COMEBACK, THE OFFENSE MUST OVERCOME A DEFICIT. Trust me; the importance of that statement justifies the usage of the caps lock. Not all comebacks are GW drives, not all GW drives are comebacks. If you never trailed in the 4th quarter, but the game is tied and you lead a drive to win the game, that is a GW drive, not a comeback. I’ll use the reigning champion 2008 Steelers as an example, considering they’re recent and six of their 19 games played were won in this fashion.
In their first meeting against the Ravens, the Steelers led 17-13 to start the 4th quarter. They added a FG, then Baltimore tied the game at 20 with a TD. In OT, the Steelers drove for the game-winning FG.
RESULT – Game-winning drive (OT), not a comeback
Against the Dallas Cowboys, the Steelers trailed 13-3 to start the 4th quarter. They added a FG to cut the deficit to 13-6. Roethlisberger then completed 4 passes for 57 yards and the tying-TD to Heath Miller. On the very next drive, Tony Romo was intercepted by Deshea Townsend for the winning TD in a 20-13 victory.
RESULT – 4th Quarter Comeback, not a game-winning drive.
In Super Bowl 43, the Steelers blew their 20-7 lead and found themselves trailing 23-20 in the final 2:30. Roethlisberger led the historic TD drive, capped off with Holmes’ game-winning catch for a 27-23 victory.
RESULT – 4th Quarter Comeback and game-winning drive.
Those are three different examples of the types of 4th quarter wins you can achieve, and with the help of boxscores, play-by-play and newspaper articles/archives, it should not be that difficult to classify them.
Checklist of questions to ask:
Did the team win the game?
- If the answer is no, then move onto the next game.
Did the QB ever have the ball in the 4th quarter or overtime with a tie or deficit of 1--8 pts?
- If the answer is no, then move onto the next game.
Did the winning team ever trail in the 4th quarter?
- If the answer is no, then this cannot be a comeback.
Did the offense produce the winning points or was it a return by the defense/special teams?
- If the answer is yes, then it’s a game-winning drive (and if there was a deficit, a comeback).
- If the answer is no, then the QB/offense does not get credit unless they did something to force a tie or get a lead at some point.
Did the offense produce a tying drive and then watched the defense/special teams score the winning points?
- If the answer is ‘yes’, then it’s a comeback, but not a game-winning drive.
All of these types of drives are positives for the offense and QB in question. Winning a game that is tied may not be as impressive as overcoming a deficit, but if you never make the plays to do it, that game may result in a loss. I think all of these kinds of drives and comebacks should be bunched together into one collection of games, and we can call them something like “4th Quarter/Overtime Wins” or “Wins Decided in the 4th Quarter/Overtime”. Then in addition to that total number, we can say how many of those wins were comebacks. That is what I did with Marino & Elway in part I. I said Marino had 51 overall wins and 36 were comebacks, while Elway had 49 & 34 (50 & 35 if you want to count that tie).
When the networks decide to show just how many comebacks/GW-drives Eli Manning has when that situation comes up in a game, they can display two numbers: 14 wins, 12 comebacks. This lets the viewer know how many comebacks he has, how many times he only had to break a tie, and overall how many times he’s come through in this drive situation they’re about to see unfold.
A popular term some teams use in their media guides are "game-saving" drives. This would be fine if everyone else was on board with it and counted games the same way. But there is no standardization and teams can basically count whatever they want. My method would provide structure. It would keep things on an even level.
Let me just state that John Elway (or any other QB) did not do anything wrong here. To the best of my knowledge, he did not instruct the Broncos to count any game they could as a comeback. The Broncos were allegedly the first team to keep track of comebacks after being asked by fans how many comebacks Elway had in the 80’s. Other teams followed suit for their star QBs, but not everyone used the same definition of a comeback. The following is a table that shows how various teams calculated comeback totals differently for some popular QBs. Using my methods to track these games, I put my actual number of comebacks up against the widely reported figure.
QB Reported Actual John Elway 47 34 Brett Favre 42 27 Dan Marino 37 36 Peyton Manning 37 28 Drew Bledsoe 32 24 Joe Montana 31 31 Johnny Unitas 31 34 Tom Brady 28 20 Roger Staubach 23 15 Ben Roethlisberger 19 15 Chad Pennington 7 7 Jay Cutler 7 5
The very first question you may ask is, "why are Chad Pennington and Jay Cutler on that list?" It’s just to show that the Dolphins and Broncos are staying true to form in their tabulations. The Broncos have no problem mentioning games that broke ties, while the Dolphins only focus on true comebacks.
Ben Roethlisberger already having 15 legit comebacks in five seasons is pretty impressive. If he can stay healthy and the Steelers continue their winning ways, he could be a threat to challenge the record holder (which is Elway, should be Marino, probably will be Peyton) some day. In the 2008 Steelers media guide, they list Ben as having 13 game-winning drives in the 4th QT/OT (12 reg. season, 1 postseason). Yet in the press release for Super Bowl 43, they say Roethlisberger had 5 during the 2008 season for a total of 17. Either they forgot to count the postseason one they had in the media guide, or they don’t want to count the postseason. With his drive against Arizona, Ben has 19 overall 4th quarter wins and 15 comebacks.
Roger Staubach was known as 'Captain Comeback', but he must have built that legacy squarely on an amazing comeback off the bench in the playoffs in 1972, and the Hail Mary to Drew Pearson against Minnesota three seasons later. He only had 15 comebacks, while every site says 23. I did locate those 23 games, and found that the Cowboys never trailed 8 times (interestingly enough 7 of the 23 games were against the Cardinals). Even the Cowboys’ official site says Staubach led 23 come-from-behind wins in the 4th quarter when they selected him the #1 Cowboy ever. My apologies to Captain Comeback and his fans, but the facts do not justify the moniker. Terry Bradshaw, a rival QB of Staubach’s, had 19 comebacks (four in the postseason). And not to stick the knife in deeper, but Troy Aikman had 16 comebacks in his career. I do not know where Danny White ranks, but Tony Romo is at 6 and counting.
I already talked about a couple Drew Bledsoe games. Now let’s look at Mr. Patriot himself, Tom Brady. Now frame this, as this will be the only time you see me defend Tom Brady. According to his Patriot bio, in the Giants victory that made New England 16-0, Brady “led the Patriots to victory after trailing in the fourth quarter for 28th time of his career.” Not true, as 9 times they never trailed. I have 29 games for Brady, but for some reason sources do not count the first game of the 2006 season against Buffalo. Trailing 17-14 to start the 4th quarter, Brady led a tying FG drive. On the ensuing drive, J.P. Losman was tackled for a safety and the Patriots won 19-17. If you’ve understood everything so far, you know that this is a 4th quarter comeback (but not a GW drive). How is this any different than one of those Elway games where Elway ties it and they return a blocked FG for a TD in overtime to win the game? It’s the same situation, yet Brady gets no credit for that game. He should have 29 overall wins and 20 comebacks.
Brett Favre is supposed to stay retired, so hopefully that will spare us any chance of the media claiming he is chasing the comeback record. He is credited with 42, but only has 27 comebacks, and is not exactly known for any real famous ones. Maybe the long game-winning TD pass to Sterling Sharpe against the Lions in the playoffs would be worth mentioning. Just not a situation Favre thrived in. The opportunities were certainly there for him to have more than anyone, but he did not come through with the record amount. A propensity for turnovers via forced throws is not what you look for in a QB in this situation.
Let’s have a round of applause for the 49ers and Chiefs for keeping it legit for Joe Montana and the 31 comebacks he made in his career. Of course it was not too hard considering he only had three other games in his career where he led a game-winning drive to break a tie. This just speaks to the dominance of the 49ers. Rarely found themselves behind in the 4th quarter, but if they were, Montana could lead them back with the best of them. Lots of memorable TD drives, and that’s what stands out about Montana. He led TD drives, several times in the playoffs, and he was usually the catalyst of the drive and the guy that threw the winning score.
According to the last press release for the Indianapolis Colts, Peyton Manning is credited with 36 game-winning drives in the 4th QT/OT. For some reason, they choose not to include the 2006 AFC Championship, which was Manning’s shining moment. Manning threw for 155 yards in the quarter, as he led the Colts to 17 points on three scoring drives in the quarter, and a 38-34 victory. They also do not include Manning’s week 16 performance against Jacksonville from last season, another of the finest games of his career. Trailing 24-14 to start the 4th quarter, Manning passed for 110 yards on two drives to tie the game at 24. On the ensuing drive, David Garrard was intercepted by Keiwan Ratliff for the game-winning TD. Obviously, it’s a 4th quarter comeback for Manning, but he doesn’t get the credit from his own team because it wasn’t a game-winning drive. That gives Manning 38 overall wins, and 28 are comebacks. He’s the favorite to take the comeback king title, especially if the Colts play games like they did last season. In 2008 alone, Manning led 4 comebacks and 3 game-winning drives in the 4th quarter.
You may remember I mentioned at the end of part I that Elway may only rank 3rd all time in comebacks. The last QB I want to talk about is Johnny Unitas. Known for crafting the 2-minute drill, Unitas led a ton of late-game rallies in his career. The only question is how many? The same Colts media guide says Unitas had 31 as a Colt. Given that he had none in his brief appearance as a Charger, it’s safe to say they’re giving him 31 for his career. And assuming they do not count postseason like they did with Manning, they are not counting "The Greatest Game Ever Played." But 32 still does not jive with what I found. Due to the fact these are older games and things can be less accurate, I am not as confident in Unitas’ data as I am that of Elway and Marino (and the other QBs mentioned). Though I will still present the case for how Unitas can be anywhere from 3rd to 1st in comebacks.
I have 43 games for Unitas. 34 are comebacks, 7 were game-winning drives that they never trailed, and 2 were comebacks he led in games that resulted in a tie. The two ties are pretty impressive for the non-OT era. Against the Lions in 1965, Unitas threw 2 TD passes to John Mackey in the 4th quarter to force a 24-24 tie. Two years later in Minnesota, Unitas twice led the Colts to tying TDs in the 4th quarter when facing a 7 point deficit. The game ended in a 20-20 tie. If you believe that ties should count, then Unitas would have 36 comebacks, the same number as Marino; the most in history.
There is another game to consider with Unitas, and it’s another one of those nasty Fiedler/Griese examples. Playing at Detroit in 1963, the Colts trailed 21-16 to start the final quarter. Unitas led a drive that resulted in a 45 yard FG to cut the deficit to 21-19. Milt Plum, throwing his first and only pass of the game, was intercepted for a TD by Andy Nelson. The extra point failed, and the Colts had a 25-21 victory. We know Unitas did not lead a game-winning drive; the defense took care of that. We know he did not lead a comeback drive to tie the game as well. The difference between this game and the Dolphins/Broncos game is that the Colts still win this game no matter if Unitas led the FG drive or not. With the other game, Fiedler’s TD pass was crucial in giving them the lead (and win). Keeping a butterfly effect in mind, does Plum ever throw that pass at that field position if Unitas never got the FG drive? I lean towards not crediting Unitas for a comeback in this situation. They needed 5, he got them 3. Could he have led another scoring drive to win the game? Of course. The defense took care of it for him though. If you are a huge Unitas fan and want to truly believe this should count (in addition to the two ties), then that would give him 37 comebacks, the most ever. But I do not think you should count this game.
Additionally, there was a game in 1970 against Buffalo that presents a rather unique case that has yet to be discussed. The game was tied at 14 to start the 4th quarter. Unitas led a go-ahead FG drive, only to see Buffalo tie it with a FG of their own. The game would end in a tie. Unitas did not lead a game-winning drive as there was no win to attach it to. He did not lead a comeback as there was never a deficit in the 4th quarter (though the Colts did trail 14-0 in the 2nd quarter before scoring 17 unanswered). This game is a positive for Unitas, given that he helped wipe out a 14 point deficit and did lead what could have been the game-winning drive, but I think in terms of 4th quarter comeback analysis, you just make a note of this one and keep it separate from the other games.
As for explaining my discrepancy with the Colts’ media guide (34 to 31), the best I can say is they must not be counting the 1958 championship game, and they probably are not counting this game against Green Bay in 1958. The Colts trailed 17-14 to start the 4th quarter. Unitas led a tying FG drive, then the defense intercepted Bart Starr for the winning TD. This is just like the Manning/Jacksonville example, which the Colts did not count. It’s a comeback, but not a game-winning drive. Subtracting those two games, it’s still 32 to 31. My best guesses are that there are errors in a boxscore(s), Unitas may have not finished one of these scoring drives, and that the Colts just simply missed a game somewhere.
Comebacks and game-winning drives are interesting subjects. On the field they make for some of the most memorable moments in NFL history. Now if everyone can get on the same page with how to classify and analyze them, it would make discussion a lot easier and more productive. No longer should someone be able to simply drop “47” and end the debate. Maybe it’s just a fantasy of mine, but I’m looking forward to the day where someone can fire back, “well he had 6 more comebacks, but on 18 more opportunities and he only led TD drives 45% of the time and had twice as many turnovers in the process!”
Feel free to send any special questions or comments about this to me at smk_42@yahoo.com.
August 7th, 2009 at 9:11 am
Nice work, Scott. I found myself nodding in solemn agreement as you described the pitfalls of counting any game where the team comes back as a comeback for the quarterback. I researched and wrote that article on Football Outsiders that you cited yesterday (thanks for the reference, by the way) and ran into more tricky scenarios than I expected. One difference for me was that I started out looking more at franchises and head coaches and how they came from behind or blew late leads. The traps you mention were irrelevant at that point. Later I assigned a comeback to whoever was the QB of record, and this led to a few difficult cases. I think my favorite non-comeback was credited to Josh McCown, who rallied the Cardinals from 1 point down in the fourth against the Seahawks in 2004 by inspiring his special teams to get a safety on a Seattle punt. I didn't discriminate in cases like that; McCown was credited with the comeback, being the QB of record. Did I feel dirty? Sure, a little.
I wish we had play-by-play data going back further in history than 1996 so I (or you, or anyone) could research the percentages of opportunities converted. Still, it was great to at least see reliable raw totals for guys like Staubach and Unitas, so again, nice job!
August 7th, 2009 at 9:35 am
Scott,
Great articles. Top-notch research & commentary.
1. Yes, it would be nice to know the opportunities and %.
2. I will say that being on the field, handing off to star RB, moving the team down for the GW score is still good. Unless you have the videotape, it is quite possible that the QB audibled out of a pass to a draw play that gained good yardage.
3. I don't think the FG vs. TD comparison is necessary. First, if you only need a FG and you kick it with less than 10 secs. left, it's just as good as the 1 yd. TD fullback plunge with 1:45 left. ESPECIALLY if that TD puts you ahead by 3 points or fewer. Grant it, you HAD TO HAVE the TD to get the points to pull ahead, but the other team now has a chance.
4. If the QB (not his team) has 3 or more turnovers in the game, that game should have an asterisk in the totals--he practically LOST the game for his team. However, if he wins a 35-34 shootout, he definitely deserves the credit, as the DEFENSE practically lost the game, not the QB.
August 7th, 2009 at 9:41 am
Very nice post and analyses. I remember when the Vikings branded Tommy Kramer as "Two Minute Tommy" and the Browns were named the "Cardiac Kids" under Brian Sipe.
.
If I wanted to conduct my own study where can I find a source available to the general public that lists all plays from games, even dating back to the 1970s and 1960s? Any help would sure be appreciated.
August 7th, 2009 at 10:00 am
hey, what do you guys have against tom brady? you always rank him real low in your lists and always seem to praise peyton manning (a guy who cant get it done in the post season). i dont think football is a stat sport like baseball is and i think its all about winning. i know tom brady had a much better team but he has played pretty darn good in superbowl games and led a comeback against the gmen in the 08 superbowl only to see his defense screw it up.
August 7th, 2009 at 10:46 am
I appreciate the hard work and also the effort to dispel myths and qualify numbers that are thrown out but not authentic. Insofar as Staubach is concerned, part of his aura was also built on two games they lost--the two SB's against Pittsburgh involved "heroic" efforts but without winning. Also, many of Dallas' victories with Staubach occurred on the big stage of nationally televised games when there was also less media exposure. Finally, Staubach's last regular season game was a two td comeback against Washington. Alas, ironically he failed in a comeback attempt against LA in the playoffs.
My point is that all kinds of "soft" factors influence one's reputation as a "clutch" QB. Also, the definition of a comeback is not universal and the article tries to "operationalize" what a comeback is. I am sure other's might disagree though with that definition. But still, before anything can be quantified, it must be precisely defined (basic to any quanititative measurement of a factor). Pardon the geek speak.
August 7th, 2009 at 11:29 am
Re: post #3b, in an earlier PFR blog one poster had revealed a pdf of the play-by-play from the 1972 Jets at Colts game, which I was able to retrieve. But when I tried to access other games I came up dry. When I contacted the site and asked for a user name and password, I was told that I needed press credentials or something like that.
I am researching something important where play-by-play info is paramount to my study. From the looks of things (post #1, last paragraph) it seems as though vintage pro football content is under lock and key?
August 7th, 2009 at 12:40 pm
It'd be interesting to ignore who won the game and just look at 4th quarter drives while behind that tie it up or take the lead, regardless of who wins. If the Broncos had lost that game in OT instead of winning, that 98 yard drive at the end of regulation would still have been fantastic.
August 7th, 2009 at 12:52 pm
I just fundamentally object to anything making QB-to-QB comparisons based on a team achievement. I know you've made some efforts to subtract out when the defense or special teams were clearly responsible for the win, but I still don't know that it tells us anything about one QB versus another, especially when some QBs are victimized by their defenses or special teams.
Consider this example: final week of the 2006 season and the Bengals are playing the Steelers with a playoff spot on the line for Cincy. Trailing 10-14 midway through the 4th, Bengals drive 73 yards for a TD to take a 17-14 lead with less than 3 minutes left. Pitt hits on two quick completions to set up a FG for the tie with under a minute left. Carson Palmer then hits Chris Henry downfield to set up a 39 yard FG...which is missed. Roethlisberger connects with Holmes for a 66-yard TD in OT for the win.
If I understand correctly, Big Ben gets credited with a comeback and a game-winning drive, Palmer gets credited with...nothing. But does that really tell us anything about the two QBs, both of whom came through with the game on the line (twice), other than one plays on a team with a lousy defense and special teams prone to very ill-timed mistakes? (I'll let everyone guess which QB I'm talking about here)
I think it would be fascinating to see something like which QBs were involved in the most 4th-quarter comeback or game-tying drives who still ended up losing, leaving out those losses when the QB was largely responsible for the loss (pick-six in OT etc).
August 7th, 2009 at 4:07 pm
Re: #1 Jason McKinley
Those weird situations are tough, but luckily there aren't too many of them. For a team that's been around since 1960, you may find about 3-5 in their history. Funny you mention Josh McCown, as the Cardinals have had quite a few of those type of wins over the years. There's a game with Jake Plummer where they trailed one time and immediately returned the kickoff for the winning TD. And people say, "comeback Jake Plummer". No, no, no. That's exactly the kind of stuff that needs to be eliminated.
3 years ago when your article came out I already had a lot of data myself on the topic and was checking my numbers with your's. To this day I still for the life of me cannot figure out how you came up with Tom Brady's failed opportunities. I always had one less. It's common for two people to look at this and come up with different numbers. That's what makes standardization so important. If this stat's going to have any meaning, it must be done right and fair for everyone. The lack of data makes that difficult.
August 7th, 2009 at 4:11 pm
Re: #3 Patrick W
Sipe and Kramer both had a pretty fair share of them for QBs not really known as franchise guys in the same way Marino & Elway are.
I think you're out of luck now for play-by-play that far back. The first season I'm aware of is 1999 (can be found on CNNSI). I know Football Outsiders has pbp back to 1996, but I've never found the 1996-98 seasons on the net myself.
Maybe some day it'll be out there. For now you can use the boxscores and follow up with newspaper archives via google searches. Use the timeline option and pinpoint the year (or even month/week) of the game you're looking at to get the most results on it. Good luck.
August 7th, 2009 at 4:17 pm
Re: #5 Tim Truemper
I think the Staubach finding is almost as good as the Elway/Marino findings. Just search around and you'll see a lot of sources claiming he had 23 "come-from-behind wins" in his career. And it's just not so.
The game against Washington in 79 is probably his 2nd best comeback behind the 49er playoff game. I'm not as big a fan of the Pearson push-off drive. But that was impressive stuff to get the Cowboys the division in his final regular season game.
The stages you do it on definitely have an impact on how people view you. But I still think the overall inconsistency in coming up with the number has a lot to do with it as well.
August 7th, 2009 at 4:25 pm
Re: #8 Ryan
That's the real heart of the analysis: examining performance. Palmer didn't do anything wrong in that game, but neither did Ben. It was a back-and-forth quarter between division rivals in week 17. You are right that Ben gets credit for a comeback/GW drive and Palmer ends up with a failed attempt.
But look at it this way. If you don't even analyze the 4th quarter, all you do is say Ben won and Palmer lost. Is it ever "fair" to give a QB a loss in a game like that when he did what he had to do multiple times? No, but we have to do it, and we always do it. Only one team can win.
This is when coming up with advanced stats and having the play-by-play really helps. I give Palmer credit for his performance in that loss. I highlight that game and make sure it stands out among his list of failed attempts.
It's not the overall number of successes and failures that mean the most, but the overall performance. We can come up with something to show how often Palmer comes through, no matter if they win or lose the game. We can do this for older QBs too, but it won't be as accurate since we don't know what they did each drive. If you get a FG one drive, then throw a pick the next, that's not good.
August 7th, 2009 at 7:19 pm
#4
Yes how DARE they put up a list of QUANTIFIABLE accomplishments and put Tom Brady lower than Manning jsut because he has fewer such accomplishments? For shame!
They should follow the SB XXVI voters example and make him MVP in the face of all objective evidence.
What's really sad is that this kind of rethoric, which you see all over football sites actualyl jsut serves to disminish Brady's acocmplishments, since you have to trot out the evidence of why he isn't Unitas/Montana/Marino squared, and thus have to focus on his shortcomings rather than his very real pluses.
August 7th, 2009 at 7:59 pm
I'm not a fan or a hater, but you have to be pretty good to throw for 4800 yards and 50 TDs even WITH Randy Moss on the team. When I wrote an algorithm to rank best passing seasons in history, it came in 12th I think. 12th is lower than most people would put it, but still, 12th is the 12th best of *all time*, pretty darned good
August 7th, 2009 at 8:15 pm
The main point I'm taking away from all this is that my man Marino has been VINDICATED, baby! Makes me proud to be a DolFan instead of a fan of one of those other teams who exaggerates the comeback numbers for their QBs
August 7th, 2009 at 8:19 pm
It also hopefully makes people more aware of just how much more Marino carried his team (and had to carry them) than other QBs of his time. I really think most people's perception of Marino is that he was just some guy who put up pretty numbers and could never come through in the clutch, when in fact the TRUTH shows he did so more times than any QB in NFL history
Seriously, those Dolphin teams in the last 80s were horrendous (last or close to last in the league in rushing and defense) and without Marino none of those teams from '86 to '89 win more than 4 games--if that many. I just wonder how many games and titles Montana or Elway would've won had they been on the Dolphins and Marino had been on their teams.
August 7th, 2009 at 11:06 pm
Scary to think how amazing Marino's stats would've been throwing to Vance Johnson and Mark Jackson instead of Clayton and Duper.
August 8th, 2009 at 1:14 am
Thanks, Scott, for the great post and replies to feedback.
August 8th, 2009 at 2:36 am
I'm trying to figure out who I can contact with the Colts to look into the Unitas numbers. They're very interesting and I think he has a legit argument to be very high on the all-time list.
August 8th, 2009 at 12:27 pm
I have been an avid Dolphins fan since the early 70’s and I have watched as many Marino games as I could over the years. In ‘98 or ‘99, I distinctly remember hearing a game announcer quote that in regular season 4th Qtr comebacks Elway had one more than Marino, but when including OT, Marino lead Elway by one game and that Marino should have the distinction as comeback king. Since his retirement Elway seemed to receive credit for an increasing number. I was never too bothered by it because he was undoubtedly a great QB. It’s funny how over the years we have become more of a society of super-hyperbole than ever before. It’s great to see someone set the record straight!
August 8th, 2009 at 3:12 pm
Since Big Ben and Cutler were looked at, I decided to look at Philip Rivers. Over his 48 NFL starts I found him leading comeback wins in eight games:
2006 Wk 10 @ CIN
2006 Wk 11 @ DEN
2006 Wk 12 V OAK
2006 Wk 16 @ SEA
2007 Wk 14 @ TEN
2008 Wk _4 @ OAK
2008 Wk 15 @ KAN
2008 Wk 16 @ TAM
The real oddity for Rivers seems to be that seven of the eight games came on the road.
August 8th, 2009 at 3:15 pm
Defining and quantifying "comeback" is great for settling claims and arguments among fans about different QBs. Thanks a lot for that.
But that said, I doubt that QB comeback wins is a particularly meaningful concept in the larger picture of things.
E.g. the claim "QB Lance Apollo won 52 games coming from behind!" implies that, well, he personally actually won 52 games for his team. His team was +52 because of him.
But football is the most "team" of all pro sports.
[] First, it's seems a lot more realistic to say his team won 52 games coming from behind, while he was playing QB. Sure, he'd have played his part in that. But his team's D stopped the other team to get the ball back with time to score (else, no comeback!) and maybe provided "gimme" field position too -- and his line blocked, his great receivers broke tackles for break-away runs ... maybe Lance himself didn't do much special at all.
Remember this very blog has reported that the difference between a starting QB and backup is about 2 points a game and just one or two wins a season -- how does that square with one player making one team +30 or +40 all by himself in the just the fourth quarter?
How many people think Tom Brady would have 20 come-from-behind wins at this point in his career if the Bengals had drafted him? So are those 20 comebacks a better measure of him personally, or of his playing on a Super Bowl contender team for his entire career?
[] Second, do we subtract from the games the QB "won" the games his team "lost from ahead" in the fourth quarter -- not just because of something horrid like his throwing a killer pick, but maybe merely because he directed a 1-2-3 out when getting a first down and using an extra minute or two would have saved the game?
Counting "comeback wins" without noting "fall from ahead losses" to judge a QB seems like judging a baseball pitcher by his total of wins, rather than W-L, or by total saves, rather than saves/blown saves.
For the record, some time back I went through Elway's games checking the results of one-score games, both wins and losses, and as I recall the W-L pct was within spitting distance of .500.
Of course that's exactly to be expected -- close games get split, because they're close! But how does that square with the idea, the legend, that Elway personally won 40+ (or 30+) close games? Because people aren't counting his close losses.
So it seems like "losses from ahead" should be subtracted from comeback wins ... But then maybe not!
Suppose a QB engineers the perfect come-from-behind drive to put his team ahead with 10 seconds to go. Then the dumbbutt bunglers on special teams or defense allow the other team to score. Does the QB not get credit for a "comeback win" performance that he certainly deserves, just because his team lost?
Soooo ... in tallying "come-from-behind wins", a mediocre QB performance carried by the rest of the team to a win gets credit for a win, a rotten performance that leads to a "fall from ahead loss" isn't subtracted from the win total, a truly great performance doesn't get any credit if it doesn't lead to a win due to somebody else's dumb fault, etc.
It seems that as a fun stat to argue about in bars it may be great and entertaining -- but as an actual measure of the quality of QB performance, and for judging HoF worthiness etc., it looks not so good.
It's not just a matter of "defining" comeback win.
There is no possible definition to eliminate the influence of all kinds of outside factors on the QB's number -- of the D in getting the ball back or not, and when and how (and even in just keeping the game close enough to be a candidate for a comeback win), the performances of all the other O players, what happens after the comeback score, etc. It's all inseparable in affecting the "comeback win total" -- and attributing the result of all of it to the QB is typical of fans, but not right.
Heck, a QB can get himself a "comeback win" by throwing three dreadful picks to blow a two-score lead in the 4th quarter, then with time running out handing the ball off to a RB who runs 60 yards against a prevent defense to score as the clock expires.
[] Third, "Why is Pennington on this list?" Favre: 27 comebacks wins in 269 games; Pennington: 7 in 77 games, which prorates to 25 in 269 games -- they're practically tied. Difference of 1 per 134 games (8 seasons). Which shows that (1) If one is going to pay attention to this stat it should be via a "per game" ratio or on a percentage of games played basis, and (2) football is a team game, results come from team play.
August 8th, 2009 at 3:20 pm
BTW That figure of 8 for Philip Rivers did not include his one playoff comeback win against the Colts on Jan. 3rd 2008.
August 8th, 2009 at 4:16 pm
Re: #20 Andy H
I debated on mentioning it and ended up leaving it out, but there was a point where Elway did have more than Marino.
When he retired after 1998, Elway had the 49 overall wins & 34 comebacks (not counting the tie). At that same point in time, Marino had 48 overall wins & 33 comebacks. By playing one more season in 1999, Marino led comebacks in Indy, San Diego and Seattle (his only road playoff win). That got his total to 51 overall wins & 36 comebacks. So it wasn't until 12/19/1999 that Marino actually surpassed Elway for good.
August 8th, 2009 at 4:28 pm
Re: #22 Jim Glass
Most of what you're saying can be applied directly to arguments on overall win-loss records for QBs. It's very similar.
When I say Marino, I think it's assumed I mean Marino, Duper, Clayton, Stephenson, Webb, Fryar, Shula, etc. are winning the game.
Goalies & pitchers still need support from their teams to win as well (unless it's a NL pitcher that can drive in his own runs and pitch a CG shutout to win a game on his own). Their records are well documented, so the same should be done for football's most important position.
It's no coincidence the QBs we rank as the best of all time usually won over 60% of their starts. Which QB's with losing records get a lot of that hype? Jurgensen? Brodie? Esiason? Pretty big drop in greatness there.
If you have a stat where about 80% of the top 10 are HOFers, that's pretty good. That's what we have here. So I think that gives it merit.
I agree that coming up with all kind of numbers & rates to identify 4th QT performance is critical, but lack of information is the setback there. Sure, we CAN do all of those things for QB's that have entered the league the last decade, but that's leaving out a lot of the greats that played in previous decades.
August 8th, 2009 at 4:49 pm
Here's my short, simpler, (cynical ?) take on "QB comeback wins":
A team is going to play a number of close games every year, basically split them, and some of those they win will be "come from behind". Then its QB will get "comeback win" credit for those, no matter who on the team deserves it or deserves to share it. (Or who on the other team deserves blame for blowing their lead.)
A QB who plays many years for a team (presumably because he's good) will get a lot of those comeback wins credited to him. And then fans will say "this great QB won all this big number of come-from-behind games for us!"
But what does it really mean?
By my tally John Elway's record in one-score games (up to 7 points differential) was 71-53, .573, +18, over 16 seasons.
And surely a good number of those +18 wins were not from "comeback" drives, but from being ahead, playing cautious and letting the other team pull within one score as time ran out.
So just what does it really mean to say, "Elway had 47 (or 34) fourth-quarter game-winning comeback drives!", when his total record in one-score games was only +18 -- and fewer than those 18 were from game-winning drives?
How does one cram 47 or 34 wins into less than 18?
By not counting what happened in the 43% of one-score games he played in that he quarterbacked to a loss.
Remember that close games are determined largely by luck, and that "great teams-coaches-players win close games" is largely a myth -- great ones win blowouts, don't lose blowouts, and split close games against other good teams.
This very blog posted the data for this on NFL coaches, showed Vince Lombardi at exactly .500 in one-score games, and gave the opinion....
"close-game winning percentage probably says more about the good or bad luck the coach experienced rather than about the coach himself".
True! And if true of the coach then it has to be true of the QB. How could it possibly be true of Lombardi but not be true of Starr? In the same game?
It follows that Elway, Marino, whoever, having a huge number X of "come from behind wins" is just a statistical artifact of their playing a whole lot of games.
It doesn't mean they weren't great players -- that's why they played so many games -- nor does it mean they didn't play great in many come from behind wins they pulled off.
But they all had a near .500 record in one-score games, as all even great teams (and thus their QBs do) ... and those close games are largely determined by luck (thus the near .500 record) ... so a huge "come from behind wins" total is basically just an artifact of playing a whole lot of games.
And as a gross number, it omits all the QB's losses in one-score games. Just as does a count of a baseball pitcher's win total without any reference to his loss total.
So "comeback wins" is a fun trivia-type stat. But it really doesn't have much at all to say about the quality of a QB's play -- except that a QB at the top of the list must have been good enough to play for a long
August 9th, 2009 at 12:37 am
One big question is whether there's much of a difference in performance in the first three quarters and performance in the fourth quarter? They guys near the top of the list were good quarterbacks all game.
One theory is that Elway has big comeback stats because his coaches tended to use him poorly until they were desperate at the end of the game.
That seems particularly true of Joe Montana in college at Notre Dame, where he led many famous comebacks such as overcoming a 22 point deficit in the Cotton Bowl and almost defeating a much better USC team. Looking back, it seems obvious that, of course, Notre Dame did best in crunch time when the coach had to let Joe Montana pass on every play than when the coach stuck to his game plan of trying to mix up the passing and running games.
After all, he's Joe Montana! But, back then, he was just some guy named Joe.
August 9th, 2009 at 12:49 am
Another question would be: how important are the skills peculiar to comebacks (clock management, quick huddle management, loud voice for shouting audibles over the roar of the crowd, aerobic fitness, cool-headedness, etc.) versus the general skills needed to succeed in the NFL?
In general, quarterbacking in the NFL is so difficult that the extra difficulty of the comeback situation isn't that much more onerous.
Here's a way to look at it: Is there anybody who had a long career who is notorious for not being able to lead comebacks?
I can think of quarterbacks failing in the clutch, like Donovan MacNabb's awful Super Bowl attempt at a winning drive, but I can't recall many teams assuming their QB would fail in the clutch and have a reliever warming up in the bullpen, the way a baseball team figures the starting pitcher should come out once the pressure mounts.
The only example I can recall of the successful repeated use of a "relief quarterback" is when the Oakland Raiders would start Darryl Lamonica and bring in aged George Blanda in the pinch.
Perhaps quarterbacks who break down in the clutch lose the respect of their teammates so fast that they are quickly out of their starting jobs.
August 9th, 2009 at 4:23 am
Jim Glass, I think the reason people don't point to records in close games is because (as you touched on) final scores are often misleading thanks to garbage time. Also with a final score you have no idea if the QB ever had an opportunity with the ball and that deficit, or if the opponent just ran out the clock. At least when you break it down by comebacks (and especially with pbp data) you know what the drive situation was and can prove that the offense still had a chance to win the game.
A lot of NFL games are close and luck usually evens itself out. I don't think luck had anything to do with close 4th quarters in Super Bowls 42 and 43. You posted the Elway numbers, and that's still over 57%. I did quick numbers on Manning, Brady and Roethlisberger and they're all over 60% while someone with a sub-.500 record like Carson Palmer still has a losing record in close games.
Something I would like to do is a regression analysis to see how good number of games won can account for number of comeback wins. You seem to think number of games started has a big impact on that judging by the fact that players like Marino, Elway, Unitas, Montana, Tarkenton, etc. all rank very high. But I disagree, because there are QBs like Archie Manning and Jeff Blake that started over 100 games in their career, but they won very few of them. They often played on terrible teams, but at many points in those 100+ starts, you have to imagine they had a chance to lead game-deciding drives and did not deliver. Dividing up the times they did and the team won, the times they did and the defense/ST still blew it, and the times they didn't would be great. No one's diagreeing with that, but you have to understand we're still held back by a lack of information to accurately judge all those things.
I also think you should consider what would become of those QBs if they didn't lead many of those drives in the 4th quarter. Marino had 155 wins as a starter counting the playoffs, and 51 of those (32.9%) weren't decided until a 4th quarter scoring drive that he led. That's a pretty significant amount of wins (and overall games in his career). If they don't finish some of those drives, they're going to lose more games, and from that, possibly miss some postseasons. That makes their room for error not as high as a QB like Steve Young. He played great on a dominant 49ers team in the 90's, and only needed 17.6% of his wins decided with a 4th quarter drive of some sort (Joe Montana - 25.6%). Does that make him a lesser QB? No, some may even argue it makes him better. But while anyone can take part in a blowout, it's usually the great ones that can overcome adversity.
Joe Montana's career would not have the same legacy if they blew the Cowboys out in 81 instead of "The Catch". He would not be known as "Joe Cool" in nearly the same degree if they blew out the Bengals instead of the John Candy moment and that TD drive.
I think something for a future part III would be to bust the myth that a 4th quarter comeback implies poor performance through 3 quarters by the QB (or any part of the team). There have been many great full-game performances that required a 4th quarter comeback for any one of several reasons.
1. NFL games are simply close - about half of them won't be decided until the 4th QT.
2. Bad defense - when you're in a back-and-forth shootout or have to make up ground your defense gave up, it doesn't mean the QB played bad.
3. All the usual screw-ups that can happen in any point of any game - skill position player fumbles, kicker misses FG's, receivers drop passes, linemen miss blocks, etc.
4. 4th quarter comebacks don't always mean Monday Night Miracle-style comebacks - I'll say it again, my hypothesis is that the comebacks that require multiple scores in the 4th quarter are more likely a result of poor QB/team play thru 3 quarters than the single-possession deficit comebacks that occur much more frequently.
August 9th, 2009 at 4:37 am
Steve Sailer, your posts bring up a few interesting facts related to Elway.
1. Elway's stats in his 34 comeback wins are better than his stats in the rest of his career games (this is not true for Marino, Manning, Brady, etc.)
2. Donovan McNabb is probably the best modern-day example of the QB that cannot get it done in this situation. It's amazing how bad their no-huddle looks while so many QB's today (Manning brothers, Brady, Roethlisberger, Palmer, Rivers, etc.) thrive in it.
3. Doug Williams had to save the day for the Redskins a few times in 1987 before taking over the starting job. And in Elway's rookie season of 1983, Elway was benched early in the season while Steve DeBerg led two 4th quarter comebacks to start the season 2-0. DeBerg threw 2 more game-winning TD passes for a total of 4 in the season. Elway led the 19-0 4th quarter comeback in Baltimore (his 1st ever, and one of the best) by throwing 3 TD passes. But early on, he saw Steve DeBerg get it done before he did for Denver.
August 9th, 2009 at 5:32 am
Now I figured out that last discrepency with the Colts media guide. I forgot to highlight the 59 championship game as a playoff game. Even though the Colts won 31-16 and it didn't carry the same luster of the 58 game, the Giants did lead 9-7 to start the 4th. Unitas ran for a 4 yard TD run and the Colts never looked back. So it's actually two playoff comebacks that he's not getting credit for, plus that regular season comeback/non-GW drive situation, hence the 34-to-31 difference.
So yeah, we have a tie for second between Elway and Unitas at 34 a piece. And if you like ties to count just the same as a win, we have a tie for first between Marino and Unitas at 36. But let's just stick with wins and a second place tie.
August 9th, 2009 at 8:38 am
how come you said you never defend tom brady? does that mean you hate him?
August 9th, 2009 at 12:39 pm
FWIW, I am piecing together Fran Tarkenton's resume on 4th QTR comebacks and have discovered a lot of what has been written about here – the “lackluster” or sub-par performance leading up to the 4th period – whether or not it was the quarterback’s fault, the obvious reliance on teammates, lacking an exhaustive list of play-by-plays, etc. My own study will take a while to finish, as I am in search of qualitative as well as quantitative information. But I have been impressed with what I see in terms of ability on Tarkenton’s part. 1972 is an interesting study, during his second stint with the Vikings in which the team posted a disappointing 7-7 record, because Tark had lots of opportunities to mount comebacks in the fourth quarter as the Vikes were not the all-world beaters they had been for the rest of the ‘70s up until Fran retired. In ’72 he led consecutive comeback drives against Miami (trailed by two in last 1:28 of the game: LOST, INT on Miami 31), St. Louis (trailed by two in last 1:38: LOST, MISSED 26-YD FG), Denver (trailed by four in last 0:53: WON with TD pass), and Chicago (trailed by three with 5:51 left: LOST, TD CALLED BACK FOR INELIGIBLE MAN DOWNFIELD, THEN 27-YD FG MISSED that would have tied it).
.
On the surface it looks as though Tark was 1-3 and any official study would tabulate that. But in grading Tarkenton he really was 2-1-1, if not 3-1 for “doing his part.” And in the loss to Miami it was revealed that Tarkenton was not calling his own plays but Jerry Burns or Bud Grant was, for which Tarkenton was upset as he had his head down each time he went over to the sidelines to discuss strategy with each time out.
.
Even in the preseason finale at Miami of that year when trailing the Dolphins by two with 0:52 left he led the Vikes into scoring position with just seconds left but saw things go for naught when Fred Cox missed a 44-yarder. Do you give the loss to Tark or Cox? A 44-yarder back then was a difficult kick. But Tark could only get them that far with the time he had left.
.
From reading accounts of the games and just from recall of having seen the events live there is no question that Tark had uncanny ability to manage the game clock, locate his receivers, make good decisions on who to throw to, and what plays to call. Often this is lost when conducting a study of fourth-quarter comebacks. But make no mistake about this, the quarterback is the most important position on the field in these situations because of all the decisions that have to be made and because of the sheer number of opportunities they have.
August 9th, 2009 at 3:37 pm
Chris, they say first impressions mean a lot on how you view people. My first impression of Brady was on a snowy night in New England against the Oakland Raiders in a playoff game. Do I need to continue?
August 9th, 2009 at 3:41 pm
Patrick, that's good stuff on Tarkenton (and really bad for the kicker). Those are the kind of things you love to find out and keep track of, but again, you never know for sure if you have all the information you need. Maybe they had three drives to get it done, but the articles only mention the last one.
The reason I don't find it criminal to give the QB a loss in that situation (though still noting the good attempt) is because ultimately, they had 60 minutes to win the game and there probably was a time in the game where they could have did something better. Kurt Warner was pretty much flawless in the 4th QT of Super Bowl 43, but his mistake was right before halftime and James Harrison made him pay with the 100 yard return. Warner led excellent 4th QT comebacks twice in Super Bowls, but he also threw a crucial pick 6 in the first half of each game that provided the winning margin.
August 10th, 2009 at 1:07 am
"Jim Glass..."
Scott, you made a lot of points, so rather than try to answer them all I'll just use an example to illustrate mine more clearly. Then I'll answer a few of your comments separately, and move on.
First, I'll repeat that I really like your work on the normalizing this statistic to make it comparable among the fans of different QBs who discuss it. That's really good, establishing a common ground to eliminate a football Tower of Babel. And I appreciate the work involved.
But as to the significance of the stat itself, I'll try to make clearer my reasons for thinking it is an artifact, in large part a result of chance-and-cherry picking, and so not really meaningful for "ranking" QBs in any serious way.
I. Take Joe Monatana: 31 "comebacks" in 164 games, that prorates to about 44 if he'd played as many games as Elway or Marino -- and that pretty much makes him "The Man" for this stat.
Montana played 13 seasons (skipping the one-gamers) and while I don't have season data on his comebacks it seems a sure bet he had 20+ of the 31 in the first 10 of his 13 seasons.
Yet his record in "close games" (7 points) in those 10 seasons by my count is exactly .500 (29-29). That's not surprising. This blog reported that Bill Walsh's record in close games (3 points) was 43%. How was his QB's record going to be much different?
So what does it mean to say: "Look how great this QB is, he has 20 come-from-behind wins in close games!" when his record in close games is .500? It means numbers are being cherry-picked, data mined. His net -20 in other close games is being ignored.
There are four "fourth quarter close game" scenarios:
1) Start ahead, stay ahead and win.
2) Start ahead, fall behind and lose.
3) Start behind, pull ahead and win.
4) Start behind, stay behind and lose.
(Plus of course teams can cycle ahead/behind/ahead...)
Say a QB has a close game record of .500. His biggest fan says,"He's got 20 come-from behind wins, proof he's the greatest!" His biggest detractor says, "The dolt has lost 20 games from ahead in the last moments, he's horrible!" They both are wrong because they both cherry picked the one number from the four that best fits their pejudice and ignored all the rest.
"Comeback wins" by itself as a stat that is inherently cherry-picked. It's just #2 alone. It's meaningless unless given with all the data for all four. It's exactly like saying "This baseball pitcher is having a great year, he has 8 wins!" How many losses does he have? ... 2?... 16?
Remember how this stat was created: The Bronco's PR department was looking for numbers to satisfy fans about how good Elway was in the last quarter. That's called data mining: "This stat? Nope. That one? Ugh. The other one? Yeah! Great!" Then they data mined him all the way up to 47!
II. It's not at all surprising that Montana was .500 after 10 years in close games because close games are mainly determined by chance. The data on this are overwhelming, for all sports. This blog reported Lombardi at 50% and Walsh at 43% in close games. It's also common sense -- it's why fans of the big-favorite team get apoplectic at their coach when the score is tied in the last moment ... they could lose due to any little thing!
As the score differential approaches zero the odds of winning approach 50% regardless of the relative strength of the teams -- see NFL sudden death overtime games.
If one accepts this fact the logical result is unavoidable:
[] Close games outcomes are determined mostly by chance.
[] Last quarter comeback games are close games.
Ergo:
[] Last quarter comeback outcomes are determined mostly by chance -- and thus so is the QB's record in them. QED.
So the number of close game outcomes a QB has reflects his ability indirectly (longevity) but his W-L % in them is mainly just chance. And for just an arbitrary subset of close games, it's even more chance!
The contrary myths that top team wins close games, and top players do too with "clutch play", just don't stand up to examination. E.g.:
"Championship teams are generally defined by their ability to dominate inferior opponents, not their ability to win close games" [FootballOutsiders]
"The #1 QB in completion percentage in the red zone last year was Sage Rosenfels. Outside the red zone however, he ranked 20th. Trent Edwards was 29th outside the red zone, but 3rd inside. Looking at the rankings, they seemed very random." [A-NFLstats] The data are endless on this...
III. Winning a high percentage of close games is NOT a good predictor of future outcomes. In fact it predicts DECLINING future outcomes. This has been known since at least Bill James's early days. Because a team that has won more than its share of close games has been lucky, its results are expected to fall (and a team that has won fewer has been unlucky so its results are expected to improve).
This is why Pythagorean rating systems based on point differentials are better at predicting game outcomes than are ELO-type systems based on past W-L.
This means that a high close-game W-L percentage for a QB -- exactly as for the team -- means he has been lucky and is not a measure of "how good" he is. In fact, it clearly systematically overstates how good he is. (There are other, much better measures of how good he is.)
So to sum up: High close-game W-L % does not show one has a superior QB or team, it shows how lucky one has been.
And "comeback wins" is just a data mined, cherry picked, perspective-lacking ("fall from ahead losses" not counted) subset of close games -- so how much is it really worth as a measure of player quality?
This is NOT TO SAY the record means nothing, or to intend any disrespect to any of the great QBs at the top of the list. To pile up these "comebacks" (like a pitcher piling up an historic number of "wins" regardless of losses) one must be a top player for many years in the league. Yes, one must be lucky too.
But one must be lucky to be on the top of any "all-time list". Babe Ruth was lucky that his home field had the shortest RF line in baseball (and that no husband ever shot him). Ted Williams was lucky to play to age 40 without getting shot down in a war. Greg Cook and Archie Manning were very good, but not lucky. Great and lucky gets one to the top tier of the Hall of Fame.
To be near the top of the "comeback wins" list one must have been a superior QB, so it shows one was one -- but it's not any measure of how superior one was to rank QBs by. It's a result of greatness, a reflection of it, a mathematical artifact of it, a consequence of it (with a lot of random noise involved) -- not a measure of it.
August 10th, 2009 at 1:43 am
Scott, replying to some of your comments:
"A lot of NFL games are close and luck usually evens itself out."
Not in a small sample size it doesn't! The NFL is a small-sample-size league, only 16 games, just 5 or 6 close games typically a season. Normal variance in that size sample produces big differences in random results from QB-to-QB.
Let's look at Montana again. After 10 years he was .500 in close games. Then in 1990 he went 9-1, +8! What ... did he and the 49ers then just suddenly learn how to win close games?
But earier in his career over a two-year stretch he went 2-10 in close games. The great Joe Montana! Look at those wild swings.
Small sample size gives high variance of results. In baseball, with 1,600 games in ten years, chance evens out. In football, no.
Montana finished in close games 43-33, 56% -- the entire +10 coming in his last three years, +8 of them in one.
~~~~
"But while anyone can take part in a blowout, it’s usually the great ones that can overcome adversity."
Exactly backwards. Anyone can win a close game with the luck of the bouncing ball or a coin-flip going into overtime. But to consistently dominate takes greatness.
Vince Lombardi and Bill Walsh didn't earn their reputations for greatness and their eight (8!) championship rings with their 50% and 43% W-L records in one-score games. They did it by destroying their opponents like nobody else did in their time.
~~~
"You seem to think number of games started has a big impact on that judging by the fact that players like Marino, Elway, Unitas, Montana, Tarkenton, etc. all rank very high."
Yup, I do, and they do. Does Bledsoe have any qualification as a great QB other than having played almost 200 games? He just had a whole lot of chances to come from behind.
"But I disagree, because there are QBs like Archie Manning and Jeff Blake that started over 100 games in their career, but they won very few of them... "
C'mon, the top seven on your list started an average of > 200 games, Blake started exactly 100, Archie 139 -- they had no chance at all for comparable numbers (even apart from being on terrible teams, and Blake being a terrible QB). You make my point.
~~
[Re close games] "You posted the Elway numbers, and that’s still over 57%. I did quick numbers on Manning, Brady and Roethlisberger and they’re all over 60% while someone with a sub-.500 record like Carson Palmer still has a losing record in close games."
1) High variance with small sample sizes -- 60% is well within the bounds of coin-flipping.
2) How close you define "close" to be makes a big difference. If you set it at 8-points a lot of your "close games" are really 2- or 3-score games -- not close at all -- in which the winning team let the other get near at the end by playing "prevent" and killing the clock.
Then good teams will be expected to have a significantly >50% record and bad ones equally far under 50%. (Brady is 60% in close games? What's the Pats W-L % in non-close games?)
OTOH, if you define close as "tied score with four or fewer minutes left" your expected result will be a lot closer to 50% -- but you will have a much smaller sample size and very wild variance from team to team (QB to QB).
Some more coaches with good sample sizes, 3-point games (from the post on this blog): Bud Grant, 54%; Shanahan, 51%; Gruden 49%, Belichick 48%, Billick 48%, Paul Brown 47%, Andy Reid 46%, Dick Vermeil 41%, Chuck Noll 40% ...
Most of those guys were pretty good and successful coaches! But up on the top of the list, coaching fewer seasons for a smaller sample: Vince Tobin 73%, Charlie Winner 70%...
Sample size makes a big difference.
~~~~
"Joe Montana’s career would not have the same legacy if they blew the Cowboys out in 81 instead of 'The Catch'."
Exactly! But if the 49ers had won by 20 because Montana had played so much better, he'd have been a better quarterback. If without the legacy of "the play".
Consider all the chance and contingency in that one play. Suppose a small bit of loose turf caused an OLman's foot to slip a little bit, throwing the play off by a tiny amount, so the pass just missed. No "legacy play". But would it have affected at all how good Montana really was? No. That's how fans misjudge players.
Think how fans pick their "clutch play heros". Yankees-RedSox tied 3-3 after the Sox come back from 0-3. Schilling with blood pouring down his leg pitching to Jeter. If Jeter slams the series-winning HR fans scream: "Great players come through in the clutch!" If Schilling mows down Jeter to win the series fans scream: "Great players come through in the clutch!" Either way another "legend" is made. Both teams are full of really great players battling each other full-tilt, but it's the team that gets the breaks that will get the credit for it and get to be legendary heros!
Now suppose one team beat the hell out of the other 4-0. No legends gets made. But there's a better team!
Bottom line: Close games and the plays in them are memorable, sure! But Bucky Dent wasn't a great all-time slugger because he hit one homer to beat the Sox in one very close game, played winner-take-all, historically remembered.
The actual measure of how good a team or QB is, is not play in close games, it's the count of butt-kicking drubbings delivered to other teams. No luck in those! Those show what "really good" is!
Though no, they don't have the dramatic enterainment value of a close game that turns so excitingly at the last minute on some little thing -- quite likely some random little thing.
Look, I know a lot of fans get really upset by the idea that luck determines close games, but it's even worse than that -- it's not just close games.
In the NFL more than 50% of all games are determined by luck.
August 10th, 2009 at 6:15 am
To be perfectly honest, I have no clue what you're trying to accomplish here.
1. Who's talking about a small sample size here? Your Elway numbers are for 124 games, which is a large sample for football.
2. Defining a game to be "close" by it's final score is not worthwhile analysis. Garbage time kills it. Not to mention you basically penalize good QBs using it that way. A bad QB will lose a higher % of games by a larger deficit, so those won't even be counted. QB's that are really good are expected to keep their team within striking distance, so a lot of their losses will be close ones. No one is good enough to come through every time and win every game, so you're going to have losses. If more than half of Montana's losses were by one score, what's that really telling you? Would it be better if he played worse and they lost by 10+ more to get his % higher in that situation? What if you're leading by 6 and get a late FG to make it a 9 pt game (and the FG basically puts the game away)? Now you're taking the "close" win away from the QB. It's a BS use of records that don't give proper credit or blame. At least with 4th quarter drives you know what the deficit and situation was and what the drive result was. That's useful.
3. I said all those QB's were over 60%, but Brady is actually 40-10 (.800) in "close" games using your +/- 7 pts definition. That's higher than his non-close game %. It also includes a loss that he didn't play past the first couple of series.
4. I don't believe anyone can make an objective system for measuring luck, so I take any of those studies with a grain of salt. That sounds like an ultimate cop-out for trying to explain why a team lost a game. So I'm supposed to believe Schilling striking out Jeter is attributed to a team getting "the breaks" rather than the pitcher being better than the batter for that matchup?
5. Sorry, but you have it wrong here. Great teams win blowouts. Great QBs should have more impact in close wins. Doesn't the fact that you keep posting these "close game" records and claiming the percentages aren't high enough prove it's harder to win a close game than a blowout? A blowout doesn't always mean the QB was great. You can fumble a ball and throw a pick 6 and be down 21-0 before you know it (sound familiar, Cowboy fans?). Or look at a team like the 2006 Bears with Rex Grossman. Lots of blowouts, but you couldn't trust the guy when it was close. At least when you need a close win, the QB USUALLY has to do something in a pressure situation to get it. Those "hand it off for 70 yard runs in the 4th QT" are way more uncommon than you think. Oh and not to mention blowout wins are often the result of playing an inferior opponent or a team that barely showed up and played lousy. So again, how is that a better measure of performance for a player than when he has to go up against a team that actually came to play and put the pressure on?
And Montana was great in all 4 SB's he played, but why is it the one that gets the most attention and the most replays is the Bengals game? Because it was the close one and he overcame adversity. He played a great game as he always did in the SB, but the difference is his team's performance wasn't as great as they were in 81, 84 or 89. That's what makes a legacy. Not blowing out a Denver team that doesn't even belong on the field with you 55-10.
6. You keep forgetting that starting games is a skill. You're not going to start 200+ games at any position unless you're good. Archie Manning could have started twice as many games, but he's still not going to sniff any comeback record because he had a losing record every single year.
7. This thought that "50% is luck and players aren't significantly better than any other players in certain situations" takes all the fun out of sports. How would you ever rank players then? Clearly some players are better than others, and they prove it on the field. If you need a 50 yard FG to win the game, try telling me with a straight face it doesn't matter if your kicker is Adam Vinatieri or Mike Vanderjagt.
August 10th, 2009 at 6:40 pm
I think trying to assess final score as a determination of QB performance in the 4th quarter is hard, as is the entire concept of "4th" quarter.
For example, consider the 2005 playoff game between Steelers and Indy. Should we ding Peyton Manning for losing a close game (21/18) when heading into the fourth quarter he was down by 18 points? Seems non-sensical
I think more important is the number of times a QB scores when trailing on the last 3 drives of the game. I would include the running plays, since a QB -can- find ways to screw up running plays as well.
From that same example, P. Manning threw a INT that the refs overturned then scored on that drive (3rd to last). The next drive he was sacked twice and threw one incomplete (2nd to last). And the last drive of the game he was 2/4 for 30 yards (Last). Clearly, Peyton should be dinged for poor play on each of the final 3 drives of the game, regardless of the final score.
I suggest looking at the number of drives out of the final 3 that an offense had a scoring drive when they were down by less than 7 or tied. If that number is 100% regardless of the outcome, I think a QB should get "full" points. If the number is less than 100%, but the QB's team win, full point. If the number is more than 0%, but the team losses, the fraction should be awarded as points.
The QB with the highest "points" divided by thier number of total games is clearly the clutch performer.
August 14th, 2009 at 3:28 pm
I have to ask about Ken Stabler. We know of two for sure: Holy Roller and Sea-of-Hands. But I know there were others. Ghost-to-the-Post might be questionable. I really am curious how many of his games under your rules would qualify.
August 14th, 2009 at 4:16 pm
Let's take a look at come-from-behind wins in golf. In major championships, Tiger Woods doesn't have many come from behind wins. But that's not because he's a choker in the clutch. It's because he's so dominant that he usually takes the lead on Friday or Saturday and cruises on Sunday.
Jack Nicklaus, I would guess, had more come-from-behind wins when he'd shoot a low number on Sunday. He also had more come from behind near misses than Tiger, sometimes when Lee Trevino or Tom Watson would stay calm and get the job done.
I think that mostly reflect that Nicklaus just wasn't quite as much better as the competition than Woods is. (I also think that Nicklaus, who tended to make a fetish out playing cautious strategies, often would dig himself in a hole on the first three days by teeing off too much with his 1-iron and then be forced to go back to his driver, with which he was great, on Sunday. Tiger doesn't make those kind of strategic errors.)
On the other hand, it does seem likely in golf that certain golfers tended to be poor Sunday players, most famously Greg Norman, although I'm not sure if buy the stereotype. Also, there are other players who appear to have been psychologically broken by blowing a lead on Sunday in a major championship, Tony Jacklin, Ed Sneed, Doug Sanders, perhaps -- and never getting back in contention after that.
August 15th, 2009 at 2:56 am
Thomas, Stabler has 26 overall 4th quarter wins and 19 were comebacks. You mentioned 3 of the most famous ones. He also had one against the Patriots in the playoffs the year Oakland won the Super Bowl. It was the infamous Ray "Sugar Bear" Hamilton roughing the passer penalty on 3rd and long that gave Oakland a first down and Stabler ran for the 1 yard TD run.
August 15th, 2009 at 4:34 pm
I'm both surprised and not. The better your team, the less likely you'll be involved in a comeback. Which would make one think that there ought to be more comebacks in the modern era than in the 70's and 80's.
If Stabler had 19, that ranks him above Staubach. Is there a somewhat complete list posted?
August 16th, 2009 at 11:17 am
It's a work in progress.
August 16th, 2009 at 7:03 pm
Steve Sailer, I think you put the jinx on Tiger. His putting was horrific on Sunday.
August 20th, 2009 at 1:58 am
"To be perfectly honest, I have no clue what you’re trying to accomplish here.
I know. Obviously. So I'll try to put it in simplest terms:
There are tons of statistics in sports. Some have predictive value, they can be used to indicate how well a player or team will perform in the future. These can be used to measure "how good", "how valuable" a player is, and "ranking" players.
A lot of other statistics are interesting and entertaining, "league records" and such, but have *no* predictive value, so they don't really show "how good" a player is except maybe in a very loose sense, so they aren't appropriate for rating or ranking how good a player is.
E.g. in baseball, OPS is a genuine measure of batting skill, carries forward consistently, and can be used to value one player against another. OTOH, "Most home runs hit in World Series play" is a stat any player would love to have and fans of that player will tout it forever. But it is near useless for ranking players, except in the very loose sense that you'd expect the player who has the record to be a very good one. He is: Mickey Mantle, 18.
But WS HRs is useless for rating players because it is hugely "polluted" by chance and contingency factors not reflecting the player's skill: 1) The number of games played in the WS (what team fate put him on) and 2) Him being "lucky", or "hot" in them. No BB player has a decent sample size of WS games. In low sample size, "seldom" events like HRs are distributed very variably, so they can't be used to measure predictible future performance.
I mean, Al Weis, an Interstate hitter, has more WS home runs that Ted Williams. Mantle having the record is an "artifact" of him being a great player, not a "measure" of it -- since it doesn't measure anything but contingency. *Some* top player would have the record. But if the trade DiMaggio-for-Williams had gone through (PR pix were taken of them in swapped uniforms) Williams might have had 20 WS HRs instead of 0. That's "contingency", what WS HRs measures.
Hey, if you are a baseball GM, who do you think has the best career ahead of him at age 27: A player with a .900 OPS and 0 WS HRs, or one with a .750 OPS 5 WS HRs. It's a no-brainer who the better player is.
"QB come from behind 4th Q wins" is an artifact record like WS HRs, loaded with contingency and chance. Sure any QB would love to have it, and his fans will brag about it. So "standardizing" the record as you have is a great service. The Hall of Fame records should be straight.
But what goes into getting the record is: #1 playing a lot of games (being healthy, sorry Namath; starting young, sorry Staubach)... #2 on a competitive team (sorry Archie). In that case the QB is going to be in a lot of close 4th quarters and win a bunch from behind, just as he loses a bunch from ahead, as a numbers game. And the count of games won (ignoring the losses, so it's not even a quality measure) is affected by the luck factor -- these wins are much "rarer" per game than "slugger" HRs, at least twice as rare, so they are much more randomly variable -- and *somebody* will get the record with a big number.
But it is no kind of meaningful measure at all of how good that QB is compared to the others. It has nary any quantifiable predictive value.
Or really, half-way through his career, due to his higher "come from behinds" count, would you prefer to have Bledsoe as your QB over Steve Young, Dan Fouts, Ken Anderson, Roger Staubach... Would you even think about it?
Bledsoe is Chase's #126! How is #126 so high on the comebacks list? Because he played a lot of games on competitive teams, a straight numbers game, with chance working in his favor (how many did he lose from ahead?) These numbers are chance and contingency.
I don’t believe anyone can make an objective system for measuring luck
Please! The first thing in stat courses is filtering out chance, reverse causation, and such. Here are the randomness factors ("signal to noise") in basic football stats. "The next chart is further illustration of the the randomness of defensive interceptions ..." Uh, oh, you don't want your GM paying a lot of cap money to generate those!
That's why it's so important to know the stats that have predictive value, versus those that don't. You want to invest in what will pay off!
"So I’m supposed to believe Schilling striking out Jeter is attributed to a team getting “the breaks” rather than the pitcher being better than the batter for that matchup?"
Geeze, a round spinning ball hitting a round swinging bat at high speed and you deny there is any chance in the outcome? Boy, you are a hard case!
Well, if one or the other is "better for that matchup" then you can know what the result will be before the pitch, right? Because if nobody can tell until after, that's pretty much the definition of chance, randomness. You can say the same thing about throwing dice.
This thought that “50% is luck and players aren’t significantly better than any other players in certain situations” takes all the fun out of sports. How would you ever rank players then?
This is a silly comment. Of course some players are better than others. Don't you want to know how and when they are more accurately, so you can rank them better? Baseball sabremetricians don't have FUN filtering all kinds of noise (park effects, etc) out of stats to rank players? They're addicted to it! Why is Chase adjusting QB ranking by weather?
Yeah, this type of analysis does explode the "fun" of a lot of old myths and legends that fans have sunk into, like belief in the "hot hand" in basketball, way overvaluation of batting average and base stealing in baseball (Maury Wills may have hurt the Dodgers). And in all sports, "great coaches/teams/players win close games" and "clutch play -- great players *improve* in the clutch!"
"Clutch play" has been sought in the data for 30 years. There's no meaningful sign of it anywhere. No clutch hitting in baseball -- sorry, Derek Jeter. In basketball, no hot hands. Do the best QBs play better in the red zone? Nope.
Exposing myths like this is progress. Yeah, it pisses off a lot of old-time fans, but progress does that. So did showing the Earth wasn't the center of the universe. Ask Galileo.
But think, the belief in "clutch play" is illogical. If a player raises his level of play in clutch situations, it means he plays less than his best at other times. What "great" player does that? What pro level athlete can get away with that? Pros who slack get cut. The pros ain't weekend softball.
There is no evidence at all for "clutch play", none, other than "memorable" plays. And memory is highly selective. When looking at the record, the "never miss clutch player" invariably has his full share of misses, forgotten from memory.
And with no clutch play, "QB come from behinds" isn't a measure of "raised-level, clutch performance", but a numbers game. (Bledsoe??)
"Clearly some players are better than others, and they prove it on the field".
Of course. But nobody suddenly becomes better than himself "in the clutch".
If you need a 50 yard FG to win the game, try telling me with a straight face it doesn’t matter if your kicker is Adam Vinatieri or Mike Vanderjagt."
And yet Belichick got rid of Vinatieri when he asked for a raise, and replaced him with an unproven rookie. Did you ever ask yourself why a smart coach like Bill would do that to "the greatest clutch kicker ever"? If clutch FG kicking is so important?
Trying to keep a straight face here ...
~~ quote ~~~
Adam Vinatieri ... is about as likely to make a clutch kick as he is to make an ordinary kick. And he is not all that more likely to make the clutch kick than the ordinarily good N.F.L. kicker. There are virtual unknowns who have a better clutch record: former Bears kicker Paul Edinger went 9 for 9, for instance. There are kickers famous for choking who were roughly as accurate in clutch situations as Vinatieri. (See Mike Vanderjagt.) As Aaron Schatz at Football Outsiders, who calculated the figures for me, says: "The sample sizes are too small to make a lot out of them. It's not really an analysis of clutch ability as it is an analysis of clutch history. And what separates Vinatieri is that he has almost half again as many attempts as any other kicker. That, and his clutch kicks are so memorable."
~~~~
"Memorable!" He was on the best team in the league, and he had more chances in big games than anyone else, so who else was going to make more "memorable" kicks?
How good was Vinatieri really? All attempts in the NFL 2003 - 2006 results rated by field position: 18th of 46. "Ordinary good." He must have had a lot of non-clutch kicks to miss!
But forget ordinary play, was he really "Mr. Can't Miss in the **Big Game**"?
Suppose your team was in the Super Bowl and you couldn't see the game for some reason. After it a buddy called you and said:
"We won by 3, but our kicker missed two FGs, from only 31 and 36."
"That choker! Can't we get a kicker who can play in a big game?"
"Well, he did make another one from all of 41"
"At least he went 1 of 3, if he'd missed that and we lost in overtime we'd have had to lynch him."
"He hit the last one with 4 seconds left."
"**HE WON THE GAME FOR US!!** WHAT A GREAT KICKER!! MR CLUTCH!! NEVER MISSES IN A BIG GAME!!"
That was Vinatieri in SB XXXVIII
Reality: If the rest of the Pats team doesn't get him a third chance to hit one, he's likely the GOAT of the game. The rest of the team does gets him a third chance, and hitting his 1st of 3 he's the HERO of the game.
Fans "remember" only the last one, forget the rest (because the team won, so they don't matter now) and he's once again "Mr Can't Miss". The story of the clutch player.
But Belichick knew the team gave him three chances to become a hero by making one ... he was 18th-of-46 quality overall ... and when he asked for a raise, got rid of him for a rookie.
As with Vinatieri so with all "clutch" players. They play at the same level all the time, like the pros they are, including in "the clutch". If they are great, yes they play great then, but the same. What's "memorable" to the fans is another story.
BTW, FootballOutsiders and AdvancedNFLStats both say place kickers are much more valuable for their ability to kick deep on kickoffs than for their FG % compared to each other. Another example of what "looks" important versus what *is* important. Most fans of course would say the opposite. Because few kickoffs are "memorable" and "clutch".
August 24th, 2009 at 7:54 pm
All I can say is good job writer. Like I mentioned in another post, I've said this for years on the ESPN message board, but not in this detail.
BTW, some websites suggest that Elway has 48 "comebacks", not 47.
And ILMAO. One website gives Favre credit for not one but two comebacks in the same game. That would be Vs the Seahawks in the 2003 playoffs. You know, where the game was won on a pick 6 in overtime, compliments "we want the ball and we're going to win".
Incidentally, if you'd like to see one of the best QB analysis' out there (actually, probably the best, and I used your stats.) then shoot me an email. You won't be sorry.
August 28th, 2009 at 12:11 pm
if the team wins with a 4th quarter comeback, whoever happens to be the quarterback gets the credit. it doesn't matter if he had a direct influence on the outcome. the only people that get credit for heroics are kickers and qbs
September 15th, 2009 at 9:34 am
I'd love to know the ratio for comeback wins compared to blown comeback wins because of a boneheaded INT. I've always argued Favre had as many if not more blown opportunities than comebacks
September 16th, 2009 at 3:58 pm
Re: #46 Jim Glass
You spent a lot of time talking about clutch play, but I never used that word. I don't believe in it in the way many sports fans do. Clutch to me is not choking, and nothing more.
About Vinatieri, well he's expensive to keep and Gostkowski was a 4th round pick and has been a very good kicker himself. As I said originally, if you think you could line up Vanderjagt for a GW kick and get the same results as Vinatieri, then good luck to ya. I've seen the games and I've seen just how wide right Vanderchoke can get them. You're talking about a guy that's choked at every level of the game from the preseason to the Pro Bowl. Yes, in that Pro Bowl game a few years back (final score was like 55-52) Manning drove Vanderjagt into FG position and he shanked it. That's a kick that's worth about what, $35,000 per player? And he missed it. For the Cowboys in their 06 preseason, he missed a game-winning FG attempt (under 40 and in a dome I believe). And here's one most wouldn't remember: in that miracle comeback the Colts had in Tampa Bay on MNF in 2003. He lined up for the winning FG late in OT and he shanked it wide right. Choked again. But thanks to the rare "leaping" penalty, he got another shot from 5 yards closer. And how'd he get it in? Doink'd off the right bar. Awful player to rely on.
While you're ripping on Vinatieri against Carolina, don't forget who kicked the 46 yard GW on a frozen night against the Titans in the AFC-D that year. In that same game, the legendary Gary Anderson (best known for choking away a SB for the 98 Vikings) fell on his arse trying to kick a 31 yd FG. Think those 3 points would have been huge in a 17-14 game? Vinatieri was also 5/5 in the snow against the Colts the following week in the AFC-C. I think he earned his ring that season. Only kicker I'd let in Canton.
September 16th, 2009 at 4:02 pm
Buck Ryan, I have Favre with 66 blown 4th quarter games (and 37 INTs & 8 lost fumbles in the process). Remember that doesn't mean Favre failed to deliver in the 4th quarter 66 times, but that's still a huge number and a ton of turnovers to back it up. Given he only has 27 comebacks and more overall wins than any QB (170 and counting), there's no doubt that Favre just wasn't very good at leading a team in a tight game in the 4th QT/OT. More of a frontrunner.
November 16th, 2009 at 1:18 pm
You can get to a pretty crude first estimate on percentage success rate by comparing career CFBWs to career losses by one score or less. The success rate is probably much higher, since many close losses would be the D giving up points the other way, but still it sets a ballpark baseline.
I came up with career stats going into 2009:
Manning 28 cfbw 18 close losses = 61%
Brady 20 cfbw 11 close losses = 65%
so far
November 24th, 2009 at 11:44 am
"Only kicker I'd let in Canton."
Which is why the Colts went and Got Venitari. And he was instrumental helping them get to the SB in 2006-2007 season.
December 26th, 2009 at 7:55 am
Would you consider this a comeback? It just happened to me on my Madden NFL 10 game.
I was down 13-20 and scored a touchdown to tie the game with 1:58 left. Then on the ensuing drive my opponent had to punt with 1:05 to go. I returned that punt for a TD to take the lead 27-20. When my opponent was trying to tie the game I picked them off and scored a touchdown to win 34-20.
My offense scored to tie the game when behind but didn't score the game winner.
December 30th, 2009 at 12:18 am
So it's a comeback, but not a game-winning drive, just like the Cowboys/Steelers example from last season.
January 16th, 2010 at 12:28 pm
Wow this stuff is great. I think the QB's job is to put his team in a position to win or score and not in a position to lose or give up easy field position or points to the other team. He cannot control whether his "team" actually wins or loses or what they doing with the scoring chances. If he puts them there then he gets credit. If he does not he gets no credit if he loses the game by eliminating a teams chance to win where they have an excellent chance to do so by doing a turnover then he gets negative credit.
There certainly can be a system designed to analyze this.
I think if losing by less than 3 then if a QB
puts a team inside the opposing teams 23 yd line in the last 3 possessions then he gets full credit for the "comeback." Partial credit can be given for 40-45 FG range. Beyond that win or lose he would get no credit. He can lose credit if he then screws up the "comeback credit" in the last two possessions. He can get additional/blame credit in same game if he puts the team again in position to win or lose. You could give less credits for position to tie. Obviously if you need a TD to win then full credit/blame can only be given if the QB scores or allows the other team to score to take the lead by pick six or fumbling.
This system would distill blame and give credit where it belongs.
Qb's should not get credit for special teams or defensive plays that get them into predictable winning range. I would say if a team gets the ball inside the opponents 23 yd line and the QB did not get them there by running or passing the ball inside the 23 yd line and they are losing by less than 3 then a win in that situation is a push. I think a point scoring system would be best.
You could actually extend this analysis to an entire game to determine how much credit a QB deserves in getting his team in and out of scoring opportunities. There will be QB's who are so efficient in the early going IE Troy Aikman that they never had to mount many comebacks in the 4th. Afterall, if you take care of business the rest of the game, you take the stress out of having to comeback.
January 31st, 2010 at 4:17 am
Just wondering if anyone knows the stats on Kenny Stabler. I remember that when I was a kid, John Madden and Stabler did the two minute drill realy well and won a lot of close games, including the unforgettable "immaculate deception" in which The Snake intentionally fumbled the ball forward and Casper (or was it Banaczak?) fell on it in the end zone against the Chargers.
February 3rd, 2010 at 10:41 am
Thanks for the fine work. I hope the NFL and the networks will recognize your contribution to the history of the game and embrace your method as a fair and valuable measure. They should christen it the "Kacsmar Formula."
In determining whether to count pre-1974 ties as comebacks, I would argue "Yes." The rules of the game in that earlier era prevented the quarterback from having the opportunity to win in overtime, so QBs should not be statistically penalized for that.
But another approach would be to subtract any OT GWDs from post-1974 quarterbacks. That would create a level playing field between the two eras.