By Scott Kacsmar
Cold Hard Football Facts’ Comeback King (@CaptainComeback)
Last week we threw down a challenge to see which single stat a person would choose to try and reflect the best possible group of the top 5 quarterbacks in NFL history.
The Captain’s answer was clear: fourth quarter comeback wins. The timing was in the anticipation of history on Monday night, and indeed we watched it happen when Peyton Manning moved past Dan Marino with his 37th comeback win, breaking the NFL record.
It sets a more defined top five, but remember the actual order was never necessarily a goal.
Most 4th Quarter Comeback Wins
Even though this is not my personal top five (80 percent close at least), trying to find a stat that does a better job is a much more difficult task than anyone expected.
We have read all of your responses and will present excerpts (sorry, some were very long and no it’s not hypocritical for me to say that in this case) from many of them as we throw out a ton of top five lists for comparison. You were told to get creative, and some of you definitely did with custom stats that no one has used before.
YOUR ANSWERS AND FEEDBACK
Will try and hit as many as possible, and section them off into different categories. Remember, there is no “correct” answer, but some are more right than others, just like how your high school English teacher graded opinionated essays.
One of the most common responses was to use a more advanced stat for fourth quarter comebacks rather than just the raw total of wins. Though in general I can agree with that, it is tough to say the raw totals are not producing a superior top five.
But Comcast SportsNet New England’s Bob Neumeier had a great idea by simply taking the number of fourth-quarter comeback wins and dividing it by games played (we included the playoffs).
However, that produces a very familiar list of five names.
If we were doing top four, then this would be my pick.
But it is the same five names as the total wins, though this is using a minimum of 150 games played. Interesting to note that Tom Brady (13.23 percent) is right behind Elway. With a top four that best suits my personal top four in order, and the prospects of adding Brady over Elway, this could be a real winner, though again it is based on a harsh restriction of 150 games.
If not, then Eli Manning (16.06 percent) would top the list and Ben Roethlisberger (15.79 percent) would be third. Also, the only 22 quarterbacks I calculated it for were the ones with at least 20 comeback wins.
Sandra Mumford echoed a similar idea, but with one slight change.
I think your comeback stat should be a percentage stat-comebacks divided by games started.
Using games started would still keep Brady out, as by my count he had two non-starts and Elway had three. It could get confusing when a quarterback has a comeback off the bench. That is not common, especially for a high-caliber player, though it’s happened at least once with Unitas.
So, perhaps a minimum number of opportunities should be considered. I
would say it takes a minimum of 4 great seasons for someone to start entering the conversation as a potential "greatest quarterback". So, assuming the average number of comeback opportunities in a season is 4 (I don't know whether this is true or not), use 4 opportunities * 4 seasons = 16 comeback opportunities as the cutoff point. This will take care of the "1 in 1" and "3 in 4" outliers. After we have the list excluding those outliers, then sort by comeback percentage (comebacks/comeback opportunities). I feel that this would give you a better list of "greatest quarterbacks.”
– Joel Cooper
The main problem with using comeback opportunities is the lack of historic data on them (though we are trying to go back). While we would love to know what Sammy Baugh’s record was in an era not known for comebacks, it is impossible right now to come up with an accurate one.
Any use of a record (percentage) would definitely have to have some type of minimum opportunity restriction, or else you end up with these absurd cases like Tim Tebow and John Skelton.
Just this week in Captain Comeback we saw that Matt Ryan (min. 30 opportunities) now has the best record in NFL history, 19-11 (.633), in total game-winning drive opportunities (a tad different than just comebacks).
If I had to make a list of the five best records I have seen (not the five actual best) in comeback opportunities with a min. 16 games, then this would be the list:
Top 5 - 4QC Opportunity % (Min. 16 Games)
Does that make for a better top five at this point than the totals, which speak to longevity? It does not appear so. One more win and Matthew Stafford (would be 8-8) moves to No. 4.
That being said, I'm a fan of the game and don't have your stats expertise but have you ever considered instead of just fourth quarter comebacks you could create a value system for the comebacks? Maybe 1pt for a comeback of 1-3 points, 2pts for a comeback of 4-7 points etc, then add up total fourth quarter comeback 'points' and take that over their fourth quarter comeback percentage for a value? Odd system but I'm just looking for a way to credit difficulty and take out the quantity factor. (I know its rough) but maybe something similar to that could help truly quantify 'clutch' better?”
– Tyler Kuehn
Tyler, there is some work started looking into using the win probability data at Advanced NFL Stats to try and quantify the difficulty aspect of comebacks (and game-winning drives). That would likely work better than any point system.
Next we have parts of a long response (over 900 words so some editing was required) from David Lopez, who is actually Rudeza Necesaria from Twitter and the person in the tweet we used in last week’s challenge article. He wanted to clarify his position.
First of all, I deny that I "do not like it when [people in Cold, Hard Football Facts] point out Aaron Rodgers’ mystifying record of 4-21 (.160) in comeback opportunities". You wrote that I disliked it while I had never said that. False statement. In fact, I acknowledge that during the "McCarthy-Rodgers era" the Packers perform badly when trailing in the fourth quarter. That is undeniable. Please note that I wrote "the Packers perform badly". These words are important. Is Aaron Rodgers to blame? In some cases yes (but not only him), in other cases not. Hey, there were cases where his performance is almost heroic.
Under McCarthy, playcalling and clock management are very poor when the game is close in the final quarter or in overtime. That is a fact. But not only with Rodgers under center. With Matt Flynn, the two Packers' fourth quarter comeback attempts (in Detroit and New England) were badly managed. There were coaching mistakes, playcalling mistakes, offensive line mistakes and receivers' mistakes. Why, then, do you attach a "W" or a "L" to the quarterback? I know, I know, for the sake of preserving the historical record, but it is truly unfair.
He goes on to state why 4QC wins are not a good stat to reflect the best quarterbacks of all time.
1. Quarterbacks do not win or lose by themselves.
Sorry, but I think that the NFL is already too "QB-centered". Thoughtful analysts (as you are, in my opinion) should contend the false (and widely spread) belief that QBs win or lose games. I do not deny that they play the most important role in a football team (more than ever in this pass-happy era), but if there is one sport where the role of every player is essential, this is football.
2. The first three quarters count too. Sorry to come back to Aaron Rodgers. His 2011 regular season may have been mediocre. At least using the “Most 4th quarter comebacks” criterion. Who played better in the 2011 regular season, Rodgers or Tebow? Rodgers or Skelton? The proposed criterion leads to blatantly stupid conclusions. Because it ignores what happens during three quarters of the game. 75% of game time is irrelevant. And quarterbacks in teams that build solid leads are penalized. Ignored.
Conclusion: I know that the names that result when using this stat include some of the greatest of all time. And I understand that Captain Comeback But they are not the greatest because of 4th quarter comebacks. They are not among the greatest because of the “stat”. This only reveals that —probably, because we have to watch the games to confirm it— they are/were reliable and valuable players for a team when trailing in the 4th quarter.
– David Lopez
David, no one is going to say John Skelton was better than Aaron Rodgers in 2011. I have never said you can just use 4QC wins to rank quarterbacks, especially in any given season.
But over the course of a career these will add up for the very best quarterbacks. The Skelton’s and Tebow’s come and go, but you will never see them rank high on the list.
And we have said this before, but should Rodgers continue to not pick up many wins in the fourth quarter/overtime, then his seasons will always be more disappointing than they should have been, and that will hold him back on the all-time best quarterback list.
Staying on Rodgers for a second, here is another Green Bay fan’s answer:
If the stat that BEST REFLECTS the greatness of a QB is quarterback comebacks, how could Aaron Rodgers have been so successful while being so pitifully bad at coming from behind? It can maybe be a deciding factor in deciding who's better between two QBs who are statistically very close, but if it is the most important, how can someone who has lost nearly ALL of 25 comeback opportunities have a .649 career winning percentage with a Super Bowl championship? I know there are anomalies in all sets of statistical data, but that one is much too large to ignore in my opinion.
I think the best indicator is still yards per pass attempt or some variant of it. It shows us how effective a passer is at moving the ball every time he drops back to pass, is less influenced by different eras, and the list of all time leaders in yards per pass attempt is incredibly similar to the list of all time leaders in a career winning percentage.
– Eric Drews
Eric, maybe Rodgers is just destined to be the ultimate anomaly. There is a reason that first article was titled “Aaron Rodgers: Front-runner Extraordinaire” last season. No one has had a career play out the way his has so far.
As for yards per attempt, it is a great stat, but Tony Romo (8.00) is still beating out Ben Roethlisberger (7.99), Steve Young (7.98) and Kurt Warner (7.95) in the top five. So that’s a no.
It sure beats yards per completion though. No one in that top five even played in the Super Bowl era. It was all about the long ball back in the day.
General Top Five Tables and Brett Favre
Here is a collection of many different tables of stats and their top five.
Pass Volume Stats
This says something about greatness being associated with touchdown passes. The top five is the same for attempts, completions, yards, but John Elway and Warren Moon drop out for Fran Tarkenton and Tom Brady in touchdown passes.
Nothing personal against Moon, but he is not a top 20 quarterback, so that eliminates these tables from any contention. In touchdown passes, that is not a bad one, but the big thing here is not having to include Brett Favre, who of course is going to dominate counting stats.
With interceptions, that top-five list includes George Blanda, John Hadl and Vinny Testaverde. Enough said.
Here is a look at key pass efficiency stats. In the NFL, you need a minimum of 1,500 passes to qualify (150 in the postseason). AAFC stats do not count.
Pass Efficiency Stats
N.Van Brocklin (8.16)
Tony Romo? Chad Pennington? PHILIP RIVERS? Way too slanted to today’s game. The ANY/A is adjusted net yards per attempt.
We can look at numbers on Pro-Football-Reference that are adjusted for era (those advanced passing tables or stats known as “Idx” in the Play Index). That should help.
- ANY/A+: Steve Young (123), Aaron Rodgers (122), Joe Montana (121), Peyton Manning (120), and Roger Staubach (120).
- Passer Rating+: Steve Young (126), Aaron Rodgers (125), Joe Montana (123), Sammy Baugh (122), and Otto Graham (122).
- YPA+: Otto Graham (124), Sid Luckman (123), Steve Young (123), Aaron Rodgers (121), and Tony Romo (120).
Well the first two aren’t bad.
More Efficiency Stats
Low Sack %
Low INT %
High TD %
The lowest sack percentage is 60 percent good, but Doug Williams and Mark Rypien? Wow. These are the only five players (min. 1,500 attempts) under 4.00 percent.
Interceptions are good for a laugh, and the high touchdown percentage again speaks to the old long-ball days of the NFL.
From e-mail, Aaron Tollerud believes most regular season wins is the best stat, not just for the top five, but for the top 11 (everyone over 100 wins). Yes, it is likely everyone in the top 11 will be in the Hall of Fame one day.
Wins as Starter
Win % (Reg.)
Surprisingly no one went for winning percentage, which is included for players with 100+ starts. Peyton Manning barely trails Terry Bradshaw right now.
The AV is Pro-Football-Reference’s Approximate Value system, which is both presented as a weighted value and the career value. The lists are the same, but with different order. They are not bad.
From e-mails, Bryan and Paul Cicero both went with playoff wins, which has Brady (16), Montana (16), Elway (14), but also includes Bradshaw (14) and Favre (13).
Honestly, the group for most playoff losses might actually be better: Favre (11), Manning (10), Marino (10), Kelly (8), and four have seven but let’s go with Montana on tie-breaker.
Favre and Montana are constant, and you are talking about Brady/Elway/Bradshaw vs. Manning/Marino/Kelly.
Another big reason why the Captain likes 4QC wins is that Brett Favre misses the cut for the top five. You see his name show up on so many of these volume lists, because no one played as many games as he did (326 including playoffs).
While Favre is a top 10 quarterback, just keeping him out of the top five is probably how most people would feel about his career. The fact that Favre is not top five – hell, the fact that he is not the leader – in 4QC wins speaks exactly to what’s wrong with him historically.
Just think of all those chances Favre had to deliver more game-winning drives, throw more touchdowns, win more games, more playoff games, and deliver great moments. Instead, we are remembering a ton of mistakes (interceptions, fumbles, sacks) in these situations.
Favre was 30-72 (.294) at comeback opportunities, and that does not even include two other games he had an opportunity that resulted in a no decision. Over 100 chances and he only produced 30 wins. That’s not good enough.
Similar with Aaron Rodgers, would we not value Favre as a better quarterback if he came through during more of those 72 losses? His stats would be better, his record would be better, his legacy would be better.
Would we not view Favre as a better quarterback if he slid for 5-8 yards or threw a great pass in the 2009 NFC Championship to set up Ryan Longwell for the game-winning field goal instead of that terrible interception to Tracy Porter across the middle? Sure we would.
There are some important (to a player’s legacy) stats no one thought of like most MVP awards, most Pro Bowls, most first-team All-Pro selections that Favre would also rank high in. Also think of the consecutive start streak.
Favre does crack the top 5 (4th) on most game-winning drives, but that’s just another reason to favor the comeback stat.
The lack of comeback wins actually speaks as loud as anything to why Favre is not deserving of a top five ranking all time. He just had too many turnovers that decided games, and not enough drives to put his team in position for a win.
Finding a stat that keeps Favre out of the top five without being too rate/new era dependent is actually a huge challenge because of the sheer length of his career. But fourth-quarter comeback wins gets the job done.
Let’s look at a collection of tweets sent on the challenge.
Ked quickly regretted his choice after checking out the leaders in lowest interception rate. Always love how Neil O’Donnell is comfortably in there even though he is only remembered for two of the worst interceptions ever in Super Bowl XXX.
Lance Newman was surprisingly the only person to mention Super Bowls won. Troy Aikman (3rd-tied) and Terry Bradshaw (1st-tied) likely scare away many people from using that one.
No one said overall championships, in which Bart Starr would have been first with five. Though, like Aikman and Bradshaw, that is probably another reason why it was not chosen.
Matt Glickman picks a good one in third-down conversion rate, though unfortunately we just do not have the historic data for that one. But it would create a pretty legit top five for active players.
Jack Cobain gets creative with a formula, though the math would not work on that one. Still, it shows what he values most.
Then you have Chase Stuart using his value system (read all about it here at his blog) to come up with his top five quarterbacks.
Chase Stuart's Value Method
It is a solid list, retaining three members of the 4QC wins list. Elway and Unitas are replaced by Steve Young and Tom Brady. Unitas came in No. 7 on Chase’s list (7,300). Elway was only 26th (4,291).
It is hard to crown any list that does not include Unitas, but this is one of the better ones out there. Joe Lellinger, via e-mail, listed a few top five tables, and included Chase’s list while also agreeing that if only Unitas was in the top five…
Some of you came up with long formulas that we do not have the time right now to calculate for every quarterback to get a top five, but it shows what aspects of the position and game people value.
For my money, Dan Marino and Payton Manning are the best throwers the game has ever seen, while Otto Graham, Joe Montana, and Tom Brady are the best leaders the game has ever seen (in that order – for now).
So in summation, my top 5 list would be determined on the following, in order of weighted importance:
1. (25% weighted) Percentage of Championship wins against seasons played (the greatest find a way to win – often)
2. (15 % weighted) Championship appearances against seasons played
3. (15 % weighted) Percentage of games where never trailing late as a percentage of games played (eliminates the longevity issue, however a 10 season minimum should apply)
4. (15 % weighted) 4th quarter comebacks as a percentage of games played (eliminates the longevity issue, however a 10 season minimum should apply)
5. (10 % weighted) Yards per attempt
6. (10 % weighted) Completion Percentage
7. (5 % weighted) Playoff win/loss average
8. (5 % weighted) Teams defensive rank by season (indicator of the teams greatness, and inversely impact a quarterbacks greatness value [the Trent Dilfer Effect]– the better the defense, the more is taken away from the overall score; the worse the defense, the more is added to the overall score – although very minimally weighted)
– Shawn Malloy
Interesting list, but there would have to be some formula combining the eight parts mathematically.
I’m not going to say I have an answer because I don’t, but in a perfect world I’d measure it something like this. Every player in every position would have a relative value; by adding these, perhaps weighting according to position, you get a value for the strength of each team relative to its opposition, and relative to the other teams of the period. Work out an algorithm which calculates the expected results of an ‘average’ QB against the known opposition, then compare those against the actual results. Your top five QBs of all time will hopefully be the best five relative values. This takes into account what they had to work with, which means you might get some very weird results.
For the record, I think your top five are pretty good, and if you’re happy with those guys, then I’d stick with them.
– Alan G Melville
Alan likes the CHFF relativity index, and wonders if it could be adjusted to reflect on the quarterback. Hear that, Kerry?
CHFF contributor Zachary Pierpoint went outside the box on his answer:
Because I was feeling bored while watching Thursday Night Football, here's a strange top 5 :
Elway, Bradshaw, Montana, Brady, Kelly
Most playoff victories, despite throwing an interception.
Elway, Favre, Bradshaw, Manning, Marino (next 5 : Montana, Kelly, Moon, Tarkenton, Brady)
Most victories (regular season or playoffs), despite throwing an interception without rushing for more than 10.0 yards per attempt
As I said, I was bored so I tried to find good top 5’s (10’s) based on "bad" statistics. In general, there are a number of statistics based around the idea of "won despite having a bad game" that make pretty good top 5’s.
– Zachary Pierpoint
“Bad stats” can still produce good lists. Helps to be on good teams, for sure.
Football researcher and friend Clark Heins keeps track of a lot of records most do not look into, and he offered his list.
As you know, my favorite stat has been road won-loss records. Only 15 QBs since 1950 have attained 10 more road wins than road losses with Montana (40), Brady (31), and P. Manning (24) being the only three topping 20 more road wins than losses. Montana and Charlie Conerly are the only two major QBs who had a better winning percentage on the road than at home.
I don’t know if this stat is any better than some other stat, but my top ten QBs of all-time are Montana, Young, Marino, Staubach, Manning, Brady, Jurgensen, Unitas, Stabler and Starr. Eight out of my top ten are on this list!
– Clark Heins
In looking at the quarterbacks with the highest road-win differential, it produces a top five of Joe Montana (40), Tom Brady (30), Peyton Manning (25), Otto Graham (19), and Charlie Conerly (19). Interesting list, Clark, which I adjusted for Brady and Manning in Week 6.
While it is clear no one should ever use one stat to make their point, hopefully what this challenge will have proven is that stats can be a very tricky beast when using them to quantify greatness.
Some stats do not transcend era well, so the old quarterbacks get the shaft. Some are hurt by compilers who hang around too long. Rate stats can favor players who did not have to play as many games, or again lean towards active players that have not yet hit their decline. Finally, the lack of data from the older days is a problem when wanting to use stats we can only calculate for active players.
As alluded to in the Manning article from Tuesday, the comeback record ages very well and has been held by a Hall of Famer all 10 times since 1950 (that’s already including Manning, a HOF lock).
Other than missing Joe Montana, who likely makes it without so many missed games in his career, the list pinpoints the best quarterback(s) from each era with incredible accuracy.
Sammy Baugh and Sid Luckman were essentially the first two franchise quarterbacks from the pre-1950 era, and they tied for the record in 1950 along with Bob Waterfield, who was part of those great Rams offenses.
Moving into the 1950s, Otto Graham, Bobby Layne and Y.A. Tittle were the best quarterbacks prior to Johnny Unitas, who revolutionized the position into what we expect today.
Unitas put the record far out of reach, and it was not until the modern era of passing and 16-game seasons that we saw John Elway tie him with 34 comeback wins, followed by Dan Marino in 1999.
Marino held that record by himself until late Monday night when Peyton Manning moved past him. Not done yet, who knows how far Manning can extend it.
When looking at the rest of the top five, Tom Brady (25) is the closest to cracking it. Even though he would replace Montana, which is a huge debate by itself, it is another case of the best of the era rising to the top.
One day it might be Eli Manning pursuing his brother for the record. You can laugh now on October 19, 2012 and say Eli will invalidate fourth quarter comebacks from producing the best top five quarterbacks, but consider the context.
Should Eli challenge 37+ comeback wins in his career, that likely means he has sustained success for another 6-9 years, which could easily mean he ends up with 3-4 rings, an epic amount of yards and touchdowns (volume stats), the longest consecutive start streak behind Favre, a ton of wins, and a ton of great moments.
People seem to be okay with Elway in the top five now, but at the Eli stage of his career (ninth season), Elway was ring-less, completed 54.7 percent of his passes, had 148 touchdowns to 140 interceptions, and a 74.4 passer rating. Top five all-time material? Hardly.
But Elway did have 20 comeback wins, or two fewer than the younger Manning right now, who still has at least 10 games left to play this season.
That is the beautifully frustrating part of top quarterback debates. Things change every year and no two people use all the same criteria.
Thanks for your feedback and perhaps we can do something like this again in the future.
Scott Kacsmar is a football writer/researcher who has contributed large quantities of data to Pro-Football-Reference.com, including the only standardized database of fourth quarter comebacks and game-winning drives. You can visit his blog for a complete writing archive. Please send any questions or comments to Scott at firstname.lastname@example.org, or you can follow him on Twitter at @CaptainComeback.