Content Post All-Time QB Rankings / QB Hall of Fame Monitor

[Ice_Ice_Brady]

The more things change, the more they stay the same.

In 1939, the Bears were riding a pretty steady franchise quarterback in Bernie Masterson, but they drafted Luckman to replace him. After Luckman's success, Halas was credited for the brilliant new system as Luckman ran the t-formation to perfection. Of course, the problem is that the t-formation was not a new concept, and things are never that seperated, where one team runs one system and other teams another system, totally different from each other.



Look at those splits...and the year after, Luckman continues his breakout and the league tries to copy the t-formation because, they think, that's the main impetus for their success...no one is saying Luckman isn't great, but the idea is that it's more about the system and concept, one that, I'll remind you, has just been sitting there, dormant, for years, in favor of the single wing.

And as teams try to copy that t-formation concept, do any sniff the level of success? Of course not. Because, as usual in the NFL, it isn't really about the Xs and Os so much but about the quarterback. Luckman has a sophomore slump season, and then in 1941, he basically breaks the league. 3 championships, and the one he lost was with arguably the greatest team ever. A passer rating of 107.5 (!) in 1943. Four first-team all-pros. Leads the league in every imaginable category.

Halas is damn good coach, but once again we see it's all about the players. After Luckman, the Bears are still looking for that great next QB, 70 years later.

 

[Deus Irae]

I love that you're still working on this, and I think it's a shame that more people haven't posted on it and offered ideas and suggestions. Great job!

 

[venecol]

I love that you're still working on this, and I think it's a shame that more people haven't posted on it and offered ideas and suggestions. Great job!

Speak for yourself. I continue to speak out for Brees, pleading for all the anti-Brees factors to be removed so he can climb back into the top 10 where he belongs, without success.

 

[Ice_Ice_Brady]

It would be a huge undertaking, but I do think it would benefit the rankings formula to include offense points scored combined with passer rating...or a better more refined way of QB measure. DVOA is good, but it only goes back to like the 1984.

Passer rating does a good job, but I think it’s heavily favoring some current QBs like Rodgers, Wilson, and Roethlisberger. They take quite a few sacks and look better on paper than Brady, Manning, or Brees, despite scoring less PPG. I could potentially use a sacks index, but I think if I were to do that, I’d have to start going deeper into adjusted net yards/attempt etc. I think some of these undertakings change the big picture results a lot less than we think they will, so I don’t know how far down the rabbit hole I want to go here.

 

[sean10mm]

I'm curious if grading super bowls based on passer rating for the game vs. the league average that season (so e.g. dead Peyton winning a ring while sucking doesn't add much) has any downsides I'm not seeing.* Like if it would bizarrely over-rate some guy who isn't good.

*Other than the work involved of course.

 

[Zarozzor]

It would be a huge undertaking, but I do think it would benefit the rankings formula to include offense points scored combined with passer rating...or a better more refined way of QB measure. DVOA is good, but it only goes back to like the 1984.

Passer rating does a good job, but I think it’s heavily favoring some current QBs like Rodgers, Wilson, and Roethlisberger. They take quite a few sacks and look better on paper than Brady, Manning, or Brees, despite scoring less PPG. I could potentially use a sacks index, but I think if I were to do that, I’d have to start going deeper into adjusted net yards/attempt etc. I think some of these undertakings change the big picture results a lot less than we think they will, so I don’t know how far down the rabbit hole I want to go here.

That’s a huge criteria of mine. I have to use PPG since DVOA only goes back to the 1980’s which is a shame. I’m hoping they can eventually at least have DVOA for post-merger.

 

[Ice_Ice_Brady]

I'm curious if grading super bowls based on passer rating for the game vs. the league average that season (so e.g. dead Peyton winning a ring while sucking doesn't add much) has any downsides I'm not seeing.* Like if it would bizarrely over-rate some guy who isn't good.

*Other than the work involved of course.

I've created an index based on how good a quarterback is overall, from a career perspective, which assigns championship shares. So, for example, Jim Plunkett isn't going to get as much credit as Roger Staubach, though they both won two Super Bowls.

In terms of going game-by-game, it's possible because I could create a score and then use it is another factor in that responsibility number.

The problem is this would have some pretty major effects on how championship points are awarded, and it doesn't account for a lot of things. For example, do you adjust according to the strength of the Super Bowl defense as well? Do you adjust for the number of passing attempts? You could also then argue about going back more rounds and look at the Conference Championship game level of play as well. Then you have other problems, too, like the NFC winning like 15 straight Super Bowls in the 80s and 90s and the game being a laughable matchup where all those QBs just padded their stats against teams that might not have been in the NFC playoffs.

I think this year's Super Bowl was a good example of how difficult it is to quantify a quarterback's contribution. This was Brady's highest rated Super Bowl...was it his best? I don't think so at all. It was his easiest, based on the built-in advantages he had with his supporting cast and defense.

For Manning, I know everyone wants to knock him down the list, but I think that not allowing guys to stack up the cumulative stats does its job of smoothing it out, moreso than taking away a few points for championships. I've tried to make the ratings so that 1 or 2 random events (like Manning winning in 2015) will not effect the scoring that much, so that even if take one away outright, it might not even change his rank, depending on how close he is to the next guy.

 

[venecol]

I'm curious what the effect of all this work has had on your original table. Is it possible to list rankings from your very first sheet with your latest? Just the ranking number and the names side by side, no other data.

 

[Ice_Ice_Brady]

I'm curious what the effect of all this work has had on your original table. Is it possible to list rankings from your very first sheet with your latest? Just the ranking number and the names side by side, no other data.

It looks like this...but the rankings should look a lot diffrent, as we're lookig at a Hall of Fame monitor versus more of an opinion on the criteria for greatness.

The list today is updated and may have some overcorrections...like Baugh moving way up and Starr dropping down. But I would just not focus on the smaller changes - those can just be a matter of a few settings...the bigger movers are interesting though....

-The counting stats guys like Favre are big fallers...the rankings I use now can use longevity, but I put it really low. Unitas is also affected by this.
-I've made some pretty big adjustments to career maturity, which is why a guy like Mahomes is a lot lower. So you have to meet a threshold where your career play no longer has a small sample size adjustment.
-Benny Friedman and Dutch Clark are the biggest movers, but that's mainly because I've been unable to properly adjust them until recently.
-Actually, Ace Parker is the biggest mover...but his career was cut short by WWII service, so I had to come up with rules that applied equally to everyone for time missed due to service, without overrewarding but without punishing for missed time either.
-Eli sucks so bad and should't be in the Hall of Fame. I've tried hard to bring him up the list, but I have nothing to work with unless I make aww-shucks, lucky ass plays worth like 7 points each.

-The reason I think the more recent rankings are a lot better than the Hall of Fame monitor is that you can focus in on an era and the intra-player rankings are still accurate, whereas with the Hall monitor, they can be very inconsistent.

 

[venecol]

It looks like this...but the rankings should look a lot diffrent, as we're lookig at a Hall of Fame monitor versus more of an opinion on the criteria for greatness.

The list today is updated and may have some overcorrections...like Baugh moving way up and Starr dropping down. But I would just not focus on the smaller changes - those can just be a matter of a few settings...the bigger movers are interesting though....

-The counting stats guys like Favre are big fallers...the rankings I use now can use longevity, but I put it really low. Unitas is also affected by this.
-I've made some pretty big adjustments to career maturity, which is why a guy like Mahomes is a lot lower. So you have to meet a threshold where your career play no longer has a small sample size adjustment.
-Benny Friedman and Dutch Clark are the biggest movers, but that's mainly because I've been unable to properly adjust them until recently.
-Actually, Ace Parker is the biggest mover...but his career was cut short by WWII service, so I had to come up with rules that applied equally to everyone for time missed due to service, without overrewarding but without punishing for missed time either.
-Eli sucks so bad and should't be in the Hall of Fame. I've tried hard to bring him up the list, but I have nothing to work with unless I make aww-shucks, lucky ass plays worth like 7 points each.

-The reason I think the more recent rankings are a lot better than the Hall of Fame monitor is that you can focus in on an era and the intra-player rankings are still accurate, whereas with the Hall monitor, they can be very inconsistent.

View attachment 31498
View attachment 31499

Wait so after all your manipulations, Brees moved up two spots? OK, good I approve.

Now please explain how the F*CK is this guy the 15th greatest QB? That looks like @Patjew's cousin.

 

[jimnance]

It looks like this...but the rankings should look a lot diffrent, as we're lookig at a Hall of Fame monitor versus more of an opinion on the criteria for greatness.

The list today is updated and may have some overcorrections...like Baugh moving way up and Starr dropping down. But I would just not focus on the smaller changes - those can just be a matter of a few settings...the bigger movers are interesting though....

-The counting stats guys like Favre are big fallers...the rankings I use now can use longevity, but I put it really low. Unitas is also affected by this.
-I've made some pretty big adjustments to career maturity, which is why a guy like Mahomes is a lot lower. So you have to meet a threshold where your career play no longer has a small sample size adjustment.
-Benny Friedman and Dutch Clark are the biggest movers, but that's mainly because I've been unable to properly adjust them until recently.
-Actually, Ace Parker is the biggest mover...but his career was cut short by WWII service, so I had to come up with rules that applied equally to everyone for time missed due to service, without overrewarding but without punishing for missed time either.
-Eli sucks so bad and should't be in the Hall of Fame. I've tried hard to bring him up the list, but I have nothing to work with unless I make aww-shucks, lucky ass plays worth like 7 points each.

-The reason I think the more recent rankings are a lot better than the Hall of Fame monitor is that you can focus in on an era and the intra-player rankings are still accurate, whereas with the Hall monitor, they can be very inconsistent.

View attachment 31498
View attachment 31499

Focusing in your top 10, 1-6 are right on.
But did you watch Starr and Unitas play? I did. They are not even remotely close. Unitas was a vastly superior wb by any measuring stick. Starr was a game manager, surrounded by the most dominant run game, dominant ol and dominant defense. Whatever input is going into your computer, your out put on this is egregiously wrong. Starr does not belong in the top 10. He's similar to Terry Bradshaw, a 15-20 guy.
Unitas was handsdown the best qb of his era.

 

[Ice_Ice_Brady]

Focusing in your top 10, 1-6 are right on.
But did you watch Starr and Unitas play? I did. They are not even remotely close. Unitas was a vastly superior wb by any measuring stick. Starr was a game manager, surrounded by the most dominant run game, dominant ol and dominant defense. Whatever input is going into your computer, your out put on this is egregiously wrong. Starr does not belong in the top 10. He's similar to Terry Bradshaw, a 15-20 guy.
Unitas was handsdown the best qb of his era.

That's good to get other opinions on this and look into why the output isn't matching up. Unitas was frequenly in the top 5 until I started putting less emphasis on counting up the numbers and looking at efficiency instead. And it was suprising that his passer rating and peak weren't close to what I thought they'd be. Intuitively, I know Unitas is too low here, and the question is how to fix it without wrecking the rest of the list.

I'd say the only big "gap" here is that situation where you have one quarterback throwing 30 TDs a year and the other throwing 15 a year, but them having similar passer rating, win pcts, etc. Usually that stuff evens out over a career, but with Unitas/Starr, it doesn't seem to. When you start moving in the volume stats, including touchdown totals, etc., it really makes things very complex and changes every measure. I'd like to have something that does that but aren't there yet...and Marino is another guy who doesn't move up the list because that's lacking.

In terms of Starr, you should read back a few pages here...Starr is one of the more polarizing figures when it comes to all-time QB rankings, with people ranking him anywhere from top-3 to lower than 15. I personalliy think he should be very high. I'd ask, for one, if you think Starr's passing numbers/passing reputation might be different if he'd had two Hall of Fame receivers like Unitas did. Not to take away from Unitas, but he was also surrounded by major talent all around.

 

[Zarozzor]

It looks like this...but the rankings should look a lot diffrent, as we're lookig at a Hall of Fame monitor versus more of an opinion on the criteria for greatness.

The list today is updated and may have some overcorrections...like Baugh moving way up and Starr dropping down. But I would just not focus on the smaller changes - those can just be a matter of a few settings...the bigger movers are interesting though....

-The counting stats guys like Favre are big fallers...the rankings I use now can use longevity, but I put it really low. Unitas is also affected by this.
-I've made some pretty big adjustments to career maturity, which is why a guy like Mahomes is a lot lower. So you have to meet a threshold where your career play no longer has a small sample size adjustment.
-Benny Friedman and Dutch Clark are the biggest movers, but that's mainly because I've been unable to properly adjust them until recently.
-Actually, Ace Parker is the biggest mover...but his career was cut short by WWII service, so I had to come up with rules that applied equally to everyone for time missed due to service, without overrewarding but without punishing for missed time either.
-Eli sucks so bad and should't be in the Hall of Fame. I've tried hard to bring him up the list, but I have nothing to work with unless I make aww-shucks, lucky ass plays worth like 7 points each.

-The reason I think the more recent rankings are a lot better than the Hall of Fame monitor is that you can focus in on an era and the intra-player rankings are still accurate, whereas with the Hall monitor, they can be very inconsistent.

View attachment 31498
View attachment 31499

Benny Friedman with the huge bump! Also interesting to see Baugh jump to #4. Is this list more based on your own opinion and evaluation than a formula?

 

[Ice_Ice_Brady]

Benny Friedman with the huge bump! Also interesting to see Baugh jump to #4. Is this list more based on your own opinion and evaluation than a formula?

Glad you noticed these two...they're connected.

So, we've been discussing for awhile about the pre-1950 stuff and assigning win % and stats with reliability. Up until now, I've had this "buffer" on all the pre-1950 players, where I've basically just faded out their scores, or regressed them to the mean, due a lack of confidence in their accuracy.

But I think I found a better way...I've been instead going through the pre-1950 players and giving them the season win/loss record based on passing attempts compared to the rest of the team. So whether or not the player is "a starter" is irrelevant to the formula...and I think that designation is screwing with the ability to get the right win %. Instead, I'm taking an approach of who is responsible for the season, and with it, the wins/losses. A lot of these QBs in the 1920s, 30s, and 40s were more of a 1-2 or 1-2-3 approach, almost like goalies in hockey. There's a clear starter, but the playing time is a lot more shared than now. So, knowing who is "the starter" is a measure we use today but wasn't always so important.



So in the end, it reduces the total games the quarterbacks get credit for, and with it, it kind of takes care of the modern/pre-modern problem by itself. That is, because the guys back then weren't 100% responsible for the passing game like they are now, they don't get as much credit for it. And while I base things on percentages, I do have a sliding career maturity scale, where less than 10 seasons does cost points...not a ton if you have like 6-7 seasons, but not full credit either, at least for winning pct and passer rating stats.

If you look at Luckman's win/loss, and you adjust it to the schedule length, he still only has about 7-8 seasons. In his case, of course, it barely even makes a ding because he's so high up, but for a lot of QBs, it's enough to push them down the list a bit.

Which brings us to Sammy Baugh, who I believe is the first quarterback in history who can account for 10 full seasons - even accounting for split playing time with other QBs - and not only that, but still boast the best era-adjusted passer rating ever. Because I was able to take off the artificial placeholder for pre-1950s, Baugh immediately shot up.



For the pre-1930s guys, I was also able to remove the constraint, and I used a formula that's converts touchdown passes to a passer rating score. It looks at the league average for that season and how many each player threw above that standard; we do have all the logs of scoring plays, so that's really the best I can do, with the understanding that with the shared responsibilities (that is also a factor) most of these scores will get watered down anyway. Friedman is an outlier there, which isn't surprising because he's an outlier in one of the seasons he played when his passer rating was actually recorded, along with the league's passer rating. So, confident that Friedman's score is consistent with his level of play, there's no reason to really to just lop off part of his score. Reading about Friedman, I believe he does belong somewhere in the top 35...20 may be high, and that ranking is on a very low longevity setting...but I don't have an issue with the system ranking him above Jim Kelly, Troy Aikman, etc.

But just to show that Friedman is an outlier, here are the other guys from that timeframe, using the same methodology...while their intra-ranking scores are interesting, there's a clear drop off on their final scores when compared to a more refined passer in the 1950s-forward. The percentage numbers next to their score show how close they are to 100%, or the standard 10 seasons of full passing responsisbilities.



In terms of my personal rankings, sometimes I'll play around with the settings to calibrate the order to my liking, but it's pretty interchangeable if you want to switch the #4 and #7 QBs or the #32 and #33, but the overal tiers are harder to change.

 

[Zarozzor]

Glad you noticed these two...they're connected.

So, we've been discussing for awhile about the pre-1950 stuff and assigning win % and stats with reliability. Up until now, I've had this "buffer" on all the pre-1950 players, where I've basically just faded out their scores, or regressed them to the mean, due a lack of confidence in their accuracy.

But I think I found a better way...I've been instead going through the pre-1950 players and giving them the season win/loss record based on passing attempts compared to the rest of the team. So whether or not the player is "a starter" is irrelevant to the formula...and I think that designation is screwing with the ability to get the right win %. Instead, I'm taking an approach of who is responsible for the season, and with it, the wins/losses. A lot of these QBs in the 1920s, 30s, and 40s were more of a 1-2 or 1-2-3 approach, almost like goalies in hockey.

View attachment 31505

So in the end, it reduces the total games the quarterbacks get credit for, and with it, it kind of takes care of the modern/pre-modern problem by itself. That is, because the guys back then weren't 100% responsible for the passing game like they are now, they don't get as much credit for it. And while I base things on percentages, I do have a sliding career maturity scale, where less than 10 seasons does cost points...not a ton if you have like 6-7 seasons, but not full credit either, at least for winning pct and passer rating stats.

If you look at Luckman's win/loss, and you adjust it to the schedule length, he still only has about 7-8 seasons. In his case, of course, it barely even makes a ding because he's so high up, but for a lot of QBs, it's enough to push them down the list a bit.

Which brings us to Sammy Baugh, who I believe is the first quarterback in history who can account for 10 full seasons - even accounting for split playing time with other QBs - and not only that, but still boast the best era-adjusted passer rating ever. Because I was able to take off the artificial placeholder for pre-1950s, Baugh immediately shot up.

View attachment 31506

For the pre-1930s guys, I was also able to remove the constraint, and I used a formula that's converts touchdown passes to a passer rating score. Friedman is an outlier there, which isn't surprising because he's an outlier in one of the seasons he played when his passer rating was actually recorded, along with the league's passer rating. So, confident that Friedman's score is consistent with his level of play, there's no reason to really to just lop off part of his score. Reading about Friedman, I believe he does belong somewhere in the top 35...20 may be high, and that ranking is on a very low longevity setting...but I don't have an issue with the system ranking him above Jim Kelly, Troy Aikman, etc.

But just to show that Friedman is an outlier, here are the other guys from that timeframe, using the same methodology...while their intra-ranking scores are interesting, there's a clear drop off on their final scores when compared to a more refined passer in the 1950s-forward. The percentage numbers next to their score show how close they are to 100%, or the standard 10 seasons of full passing responsisbilities.

View attachment 31507

I like your method to try and figure out the older players win/loss. Definitely can see why Baugh jumped Luckman now. You’ve made me research Friedman a bit more and when I went to his Pro-Football-Reference page I was shocked at the lack of data for him. But from everything I can find about him he was elite.

 

[Ice_Ice_Brady]

I like your method to try and figure out the older players win/loss. Definitely can see why Baugh jumped Luckman now. You’ve made me research Friedman a bit more and when I went to his Pro-Football-Reference page I was shocked at the lack of data for him. But from everything I can find about him he was elite.

When you go to the team pages, for each season he played, you can find the scoring logs. It tells you how many touchdown passes he threw for. In column E, I use his touchdowns/team touchdowns to determine responsibility, until 1932, when passing attempts can give a more precise measure of responsibility and passer rating gives a more precise measure of performance.





In column S, I'm comparing Friedman's touchdown passes (Column E) against league average into a percentage...I do the same thing with passer rating...so this is a little different way of doing it, but it's the only data we have since we don't have completions, attempts, ints, etc.

Friedman scores a ridiculous .864 total pct. Now, I might think that's a big error there, but it's actually similar to his converted passer rating score in 1933 against the league average (the exact same formula I'm using for every other quarterback.)

Also, using that same methodologies, here are the percentages from other players...normal converted percentages that mesh with players from other eras, in terms of matching up with their winning pct (bottom right, highlighted).







So in the end, Friedman's 1933 era-adjusted passer rating caused me to think something was wrong with the numbers. I then tried to standardize pre-1930 players based on all the info we have, and I think that their scores are very much in alignment with what we see from other players from other eras...except Friedman's pct is almost exactly the same as his 1933 era-adjusted rating that made me look into this to begin with.

 

[1960Pats]

View attachment 31468

Great stuff.

I seem to have a vague memory from way back that a pass wasn't a pass unless it was thrown overhand. If a ball was thrown forward but underhand it was considered a lateral. I can't seem to find anything about that online.

 

[Ice_Ice_Brady]

I think @jimnance brings up a good point.

I’m wondering if peak score here isn’t doing what’s intended. I’m already assigning an era adjusted passer rating, and then this is just another era adjusted passer rating, but based on best season. Peak score hasn’t really made the difference I was hoping for.

I’m wondering it makes sense to make the peak score more volume based. Maybe something like Touchdowns minus Interceptions or total Touchdowns.

I think it’s important that Marino’s peak season had 48 TDs while Griese’s had 19. Those two seasons were not as alike as passer rating would indicate. I really like Starr, but there are similar concerns about assigning him a higher peak score than Unitas...it seems to miss something important about these guys. Perhaps some volume measure is necessary here.

 

[pazrul72]

I think @jimnance brings up a good point.

I’m wondering if peak score here isn’t doing what’s intended. I’m already assigning an era adjusted passer rating, and then this is just another era adjusted passer rating, but based on best season. Peak score hasn’t really made the difference I was hoping for.

I’m wondering it makes sense to make the peak score more volume based. Maybe something like Touchdowns minus Interceptions or total Touchdowns.

I think it’s important that Marino’s peak season had 48 TDs while Griese’s had 19. Those two seasons were not as alike as passer rating would indicate. I really like Starr, but there are similar concerns about assigning him a higher peak score than Unitas...it seems to miss something important about these guys. Perhaps some volume measure is necessary here.

First time comment long time reader this is an amazing thread with a stupid amount of work put in so I thank you for that. Not sure how you could add this in but the biggest thing I think of when considering a “peak” season, Marino’s being the biggest one, is by how much did they destroy the previous record. At the time 48 TDs was insane and took 30 years to beat. The most TDs before that was 36 so a +12 jump is far more impressive then say Manning beating it by 1. Marino would get a double bonus for this one as 5,000 yards was also unthinkable at the time. Only 2-3 QBs passed for 4K a year not the 10-12 that do it every year now.

 

[Ice_Ice_Brady]

First time comment long time reader this is an amazing thread with a stupid amount of work put in so I thank you for that. Not sure how you could add this in but the biggest thing I think of when considering a “peak” season, Marino’s being the biggest one, is by how much did they destroy the previous record. At the time 48 TDs was insane and took 30 years to beat. The most TDs before that was 36 so a +12 jump is far more impressive then say Manning beating it by 1. Marino would get a double bonus for this one as 5,000 yards was also unthinkable at the time. Only 2-3 QBs passed for 4K a year not the 10-12 that do it every year now.

For sure, that would be a good measurement tool. I've tried to give more points for guys who broke the ceiling for passer rating because they've contributed more to the evolution of the position. Having these monster seasons and setting new levels of excellence for guys like Baugh, Luckman, Graham, Unitas, Marino should have have more reward.

I was really surprised that when I just put in the peak rating score, Brady came up #1 post-1967. I though there must be some error with the formula, but it turns out that the average touchdown passes per team was about the same in 1984 as 2007.