Brad Oremland is a longtime commenter and a fellow football historian. Brad is also a senior NFL writer at Sports Central. There are few who have given as much thought to the history of quarterbacks and quarterback ranking systems as Brad has over the years. What follows is Brad’s latest work on quarterback statistical production.

---

This is part four of a five-part series. It is a supplement to my 2015 series on the greatest quarterbacks of all time, last year’s article on the top-ranked QBs in Total Statistical Production, and last month’s post about QB-TSP in the 2017 season and another way of using TSP. I strongly encourage you to read those pieces if you haven’t done so already. You may also be interested in parts one, two, and three of this series.

In this series, I present the top 100 pro football quarterbacks as ranked by QB-TSP. This is a purely statistical ranking, with all the drawbacks that entails, and in many places it is not reflective of my subjective evaluations. Each week, we’ll examine 20 players, continuing this week with ranks 21-40.

40. Len Dawson — 14,225 — 17.6 — 6 – 27 – 162

At this point in my career as an analyst and historian, I am confident that Joe Namath was better than Len Dawson. This was taken for granted during and immediately after their careers, but it is practically heresy today. I understand why Dawson is commonly rated ahead, but I hope I can show why I’m so convinced that today’s conventional wisdom is wrong.

Let’s begin with stats. Dawson is far more highly regarded because his passing stats look much more efficient than Namath’s. We’ll put that in context momentarily, but here are the raw numbers:

Player  Att    Yds     TD   INT  Rating RshYd  RshTD
Dawson  3,741  28,711  239  183  82.6   1,293  9
Namath  3,762  27,663  173  220  65.5     140  7

It’s a runaway for Dawson! He’s ahead in every category. I totally understand why people look at these statistics and assume Dawson must rate not only ahead of Namath, but far ahead.

Quarterbacks are known for different strengths. Johnny Unitas and Jim Kelly were renowned for their toughness. Terry Bradshaw and John Elway were remarkable for their arm strength. No one was cooler under pressure than Joe Montana, and so on. What was Joe Namath most famous for? His quick release. Not even the most partisan Len Dawson fan would claim he could get rid of the ball as quickly as Namath. That helped Joe Willie avoid sacks and fumbles. These stats get left out of the rating formula:

Player   Sacks  Yds     Fmbl
Dawson   382    3,436   85
Namath   171    1,629   33

Sacks and sack yardage are estimated prior to 1967 (inferred from official team totals), but the difference is dramatic: 211 wasted plays, 1,807 lost yards, 52 fumbles. Namath averaged 6.62 net yards per attempt, half a yard ahead of Dawson (6.13). Half a yard per play, over 4,000-play careers … well, that’s a big deal.

Equally big: the context in which these stats were produced. Dawson washed out of the NFL. The first overall pick in the 1957 draft, Dawson attempted 45 passes in five years before the league gave up on him. Legendary coach Paul Brown, who had Dawson in 1960 and ’61, was happy to release Dawson but cautioned the Chiefs against signing him. Dawson, though — like many NFL wash-outs — thrived in the upstart American Football League. His first season playing against the weaker competition, Dawson was first-team All-AFL.

But how meaningful was that, really, in the early AFL? A year earlier, fellow NFL cast-off George Blanda was even more dominant. The following season, Tobin Rote returned from the Canadian Football League to win first-team All-AFL honors, the first of three straight CFL veterans named All-AFL QB. The AFL didn’t produce an all-league QB who hadn’t flunked out of the big-boy league until 1967. ^[1]Here’s where the first seven All-AFL QBs were in 1959: Jack Kemp — CFL; Blanda — out of football; Dawson — 7 pass attempts; Rote — 26.8 passer rating, 14.9 sack … Continue reading

Dawson had three of his best seasons in the early AFL, while Joe Namath was still at Alabama. Dawson was first-team All-AFL in 1962 and 1966, when the AFL clearly had not caught up to its older brother. Namath played in the AFL, too, but only for its last five years, and Namath, unlike Dawson, was All-Pro in the NFL, too, in 1972. ^[2]The Associated Press named Namath to the second team, behind Miami’s Earl Morrall, but every other major organization named Namath first-team All-Pro. Namath led in QB-TSP. So Dawson’s superior passing stats are almost entirely a product of [1] the weakness of the early AFL, and [2] a difference in style of play, wherein Namath threw the ball away under pressure, leading to low completion percentage and sometimes to interceptions, but Dawson ate sacks under pressure, leading to low NY/A and sometimes to fumbles.

Furthermore, Namath had big years, including five 1,500-TSP seasons and three 1,950-TSP seasons. Dawson had three 1,500-TSP seasons and no seasons of at least 1,900 TSP. He was more of a compiler, with four seasons between 500-825 TSP and another five between 1,100-1,475. Those nine seasons account for 54% of his Career Value. Namath, by contrast, draws 83% of his CV from his five very good seasons. His career stats are dragged down by three real bad years at the end of his career (22 TD, 49 INT, 47.7 rating, 4.9 NY/A). I judge players mostly by their primes: Namath’s was exceptional and Dawson’s was not. It’s to Dawson’s credit that he maintained effectiveness longer than Namath, but I hold it more against Namath’s coaches than against the QB that he was still playing in 1976 and ’77. If Namath had retired three years earlier, would that make him a better player? That doesn’t make sense to me.

When the all-time AFL team was chosen in 1970, Namath was a near-unanimous choice at quarterback (Dawson received only one vote). Namath, who retired two years after Dawson, was inducted into the Pro Football Hall of Fame two years earlier.

Unless you believe the early AFL was just as strong as the later AFL and the post-merger NFL, or you judge QBs mostly by passer rating, I don’t understand how a thorough analysis would conclude that Len Dawson was as good as Joe Namath.

39. Rich Gannon — 12,258 — 17.68 — 4 – 30 – 120

Since the 1970 AFL-NFL merger, 11 quarterbacks have earned Associated Press first-team All-Pro honors more than once: Bob Griese, Dan Fouts, Dan Marino, Joe Montana, Steve Young, Brett Favre, Kurt Warner, Peyton Manning, Tom Brady, Aaron Rodgers, and Rich Gannon. That’s exceptionally fine company for Gannon. He would probably be a Hall of Famer, except that he didn’t start regularly until he was 32.

38. Daryle Lamonica — 10,940 — 17.73 — 5 – 38 – 190

Was Daryle Lamonica a great quarterback, or a product of the teams he played on?

During Lamonica’s first four years, in Buffalo, he failed to outplay Jack Kemp. Serving as backup, Lamonica compiled a 55.0 rating on 353 pass attempts, with 6.1 NY/A, plus 8 rushing TDs. Taken together, those four years earned him 0.7 Career Value, which seems fair for a backup with a 55.0 passer rating in a weak major league.

Lamonica started in Oakland from 1967-72, before losing his job to Ken Stabler, and his case for greatness rests on those six seasons, maybe just five seasons, since he was not effective in ’71. The Raiders improved immediately when they acquired Lamonica, from a pair of respectable 8-5-1 seasons to an AFL-best 13-1, followed by back-to-back 12-win seasons. It is possible that this improvement should be largely attributed to Lamonica, but the Raiders acquired him at the same time as Hall of Famers Gene Upshaw, Willie Brown, and George Blanda, not to mention one of the greatest deep threats of all time, Warren Wells. Lamonica surely made a positive difference, but he had a lot of help.

The Raiders in that era were stacked with Hall of Fame talent. The 1971 Raiders, who finished 8-4-2 and for whom Lamonica accrued a 66.8 passer rating and 5.9 NY/A, had eight Hall of Famers, nine if you count Stabler. ^[3]That includes four of the five starting offensive linemen, plus wide receiver Fred Biletnikoff. Only the Vince Lombardi Packers had more HOF talent. There was a feeling, not only around the league but within the locker room, that Lamonica was holding the team back. Raider biographer Peter Richmond quotes an anonymous Raider: “Kenny should have been the quarterback as early as 1971. There was a constant sense on the team that Kenny should have been playing.” (Badasses: The Legend of Snake, Foo, Dr. Death, and John Madden’s Oakland Raiders, HarperCollins, 2011)

Lamonica had multiple 2,000-TSP seasons, and he was a two-time AFL MVP. But he only started for six years, and he played behind a Hall of Fame line, with Hall of Fame receiving talent, on teams that were pretty good before he arrived and consistently excellent after he left (eight straight winning seasons, including two Super Bowl victories). He was a good player, but he had more in common with Trent Green than with Roger Staubach.

37. Troy Aikman — 13,375 — 17.8 — 5 – 23 – 115

Among the Top 100 QBs in this project, Troy Aikman ranks 98th in touchdown percentage (3.50%), ahead of only Archie Manning and Kerry Collins. This is partly due to Emmitt Smith, one of the greatest short-yardage rushers of all time. But it’s not just short TDs that Aikman wasn’t picking up. From 1989-2000, a time frame cherry-picked to favor Aikman, he ranks 13th in TDs of more than 10 yards. His performance in other time frames is even more ghastly; during his prime years, from 1991-96, when Aikman made the Pro Bowl every year, he ranked 16th in TDs of more than 10 yards. His total is less than half of Jim Kelly’s. He’s behind Neil O’Donnell and Scott Mitchell. It’s hard to blame that on Emmitt.

36. Donovan McNabb — 13,717 — 18.2 — 5 – 22 – 110

There are multiple reasons that this ranking underrates Donovan McNabb, but let’s focus on just one: partial seasons. Probably the greatest weakness of TSP is its failure to recognize excellent play in partial seasons. This relates to another weakness, the Bob Griese Problem, wherein good QBs on successful run-oriented offenses are penalized for their lack of volume. ^[4]These problems are fixable; every statistical problem in TSP is fixable in a vacuum. To more fairly rate high-efficiency, low-volume seasons, simply increase the penalty for attempts and sacks. If I … Continue reading

There are two McNabb seasons that merit particular attention: 2002 and 2006. In both seasons, McNabb played only 10 games, about 60% of the season. In 2002, he passed for 17 TDs and only 6 INT, plus he rushed for 460 yards, with a 7.3 average and another 6 TDs. In a full season, he would have been an MVP candidate. For that heroic performance (with Todd Pinkston and James Thrash as his leading receivers), McNabb earned 1,172 TSP and 1.3 Career Value. His 16-game pace, which still underrates him — not only for his poor supporting cast but because TSP slightly underrates rushing — was 1,876, which would have garnered 3.2 CV, a massive 146% increase. McNabb didn’t play the full 16 games, and it’s appropriate for a statistical analysis to note that, but in this case, the penalty seems excessively punitive. He played 60% of the season and got 40% of the credit.

The same applies in 2006, when McNabb passed for 18 TD and 6 INT, with a career-best 7.44 NY/A and the lowest fumble rate of his career. Now throwing to Reggie Brown and Donte’ Stallworth, McNabb managed 1,620 TSP and 2.4 CV in just 9½ games. Projected to 16 games, he could have earned 6.4 CV, by far his best season.

Most QBs are affected to some extent by TSP’s harsh treatment of partial seasons, but McNabb is the only quarterback, certainly the only modern quarterback, to have two of his three best seasons so affected. I’m not suggesting we restore full credit for these seasons and apply a +5.9 bonus to McNabb’s CV (which would catapult him to 22nd, absent any consideration of other reasons he is underrated), but I think subjective adjustment is called for, just as it is in the case of the Bob Griese Problem, the Archie Manning Problem, and other statistically undocumented hardships or pleasantries.

McNabb made six Pro Bowls. He is one of 18 QBs to qualify for at least five Pro Bowls since the AFL-NFL merger. Here are the others: Troy Aikman, Tom Brady, Drew Brees, John Elway, Brett Favre, Dan Fouts, Bob Griese, Jim Kelly, Peyton Manning, Dan Marino, Joe Montana, Warren Moon, Philip Rivers, Aaron Rodgers, Ben Roethlisberger, Roger Staubach, and Steve Young. Other than McNabb, every one of those quarterbacks was a first-ballot Hall of Famer or is not yet eligible. In three years on the HOF ballot, McNabb has never advanced beyond the first round of voting, not even to the round of 25. I’d estimate the chances that McNabb has been fairly evaluated, apart from his race, at something very close to 0%. He’s underrated by TSP, but that’s a statistical accident. He has been systematically underrated by a majority of the mainstream sports media.

35. Bert Jones — 10,551 — 18.4 — 4 – 28 – 112

Bertram Hays Jones fumbled less often than any QB in history. His fumble rate is totally unique and preposterously low. The most predictive variable correlated with fumbles is sacks taken. There are other variables that matter as well, but in general quarterbacks fumble approximately once for every four times they get sacked. That does not mean QBs fumble on one out of every four sacks — but they fumble snaps and sneaks and everything else in a way that lines up pretty well with that figure. Among the top 100 QBs, both the mean and median are 23.6 fumbles per 100 sacks, with a standard deviation of 8.8; 77 of the top 100 are between 16.0 – 29.5. Here are the five best rates:

1. Bert Jones — 232 sacks, 16 fumbles, 6.9 / 100
2. Joe Theismann — 340 sacks, 42 fumbles, 12.4 / 100
3. Greg Landry — 316 sacks, 41 fumbles, 13.0 / 100
4. Ken Anderson — 398 sacks, 52 fumbles, 13.1 / 100
5. Fran Tarkenton — 592 sacks, 84 fumbles, 14.2 / 100

Jones is a colossal outlier. This is trivia more than significant data, and it’s tilted to favor players who got sacked a lot, which Jones did, but his low fumble rate is a massive, titanic, unearthly outlier. If Jones had fumbled at a slightly more normal rate — I doubled his fumbles every season, which would still make him 4th-best among the top 100 — he would drop from 35th to 41st in Career Value. For a fuller sense of Jones’ accomplishments and shortcomings, I’ll refer readers to my short essay on Jones for the Best QBs in History series.

34. Tony Romo — 13,052 — 18.7 — 7 – 31 – 217

The Cowboys are the only team with a primary claim on six QBs ranked in the top 100. ^[5]The Browns, Packers, and 49ers are tied for second, with five each. In a comment on my list of each team’s best QBs in franchise history, Tim Truemper pointed out that all of the top five QBs in Dallas Cowboys history played exclusively for Dallas: Don Meredith (1960-68), Roger Staubach (1969-79), Danny White (1976-88), Troy Aikman (1989-2000), and Tony Romo (2003-16). No other franchise has an exclusive claim on more than three of its top five QBs (Dolphins, Giants). The Eagles, Jets, Raiders, Saints, and Vikings don’t have any.

Craig Morton, who played for the Cowboys from 1965-74 and earned the majority of his TSP (6,039) and CV (7.2) for them, is the sixth Cowboy in the Top 100.

On this list, Romo is the lowest-ranked player with at least 200 Year-Points. He only started for nine seasons, but he was consistently a top-10 QB when healthy.

33. Randall Cunningham — 12,557 — 18.8 — 5 – 26 – 130

Perhaps the most common mistake in evaluating quarterbacks is to judge them exclusively by their passing stats. I have actually seen fans pronounce judgment on exceptional runners like Johnny Lujack, Tobin Rote, Doug Flutie, and Cam Newton without considering their rushing. That’s incredible, in the most literal sense. A similar mistake applies to players with exceptionally low sack and fumble rates: these critical statistics get left out of many analyses, leading to flawed conclusions.

Randall Cunningham, of course, is among the greatest running quarterbacks of all time, ^[6]He led all QBs in rushing six times, which is the all-time record (tied with Rote, who played in a much smaller league). probably the greatest until Michael Vick, and Vick wasn’t a fraction of the passer Cunningham was. From 1986-90, Cunningham rushed for over 500 yards every year, over 6.0 yards per attempt, and a total of 23 touchdowns. He was injured off and on for the rest of his career but continued to run effectively. In the legendary video game Tecmo Super Bowl, QB Eagles (based on Cunningham) was rated for the same maximum speed as elite wide receivers like Henry Ellard and Tim Brown. Sports Illustrated called him — Cunningham, not QB Eagles — “The Ultimate Weapon.”

Cunningham was an excellent passer, who overcame coach Buddy Ryan’s Dark Age approach to offense, posting a +73 TD/INT differential, better than contemporaries like Jim Kelly (+62), Warren Moon (+58), and Troy Aikman (+24), and basically the same as John Elway (+74). But to evaluate Cunningham without considering the titanic impact of his athleticism and rushing is nonsense, and not a meaningful or even credible analysis. While Cunningham was a unique athlete and an extreme example, a similar disclaimer applies to all talented rushers, and also to QBs who excel in other areas not included in the passer rating formula, like avoiding sacks and fumbles.

32. Joe Namath — 13,253 — 19.6 — 6 – 39 – 234

In 2018, the notion persists that sacks are somethat that happen to the quarterback, outside of his control and the exclusive fault of his blockers. Joe Namath, along with Peyton Manning, Dan Marino, and Norm Van Brocklin, were probably the greatest modern QBs at avoiding sacks.

Manning’s career sack rate of 3.13% percent is the best of all time. ^[7]Though you have to get to the fourth decimal (or sixth, depending on how you think about it) to rank him ahead of Marino. Manning had the lowest sack rate, among qualified passers, in 1999, 2002, 2003, 2006, 2009, 2010, 2013, and 2014. The 1999 Colts and 2010 Colts had no players or coaches, besides Jeff Saturday and Manning, in common. ^[8]Saturday only started 2 games in 1999, so it’s tough to give him much credit for preventing sacks. When Manning moved to Denver, he led the league in lowest sack percentage in two of his three full seasons, ranking 2nd the other year. This pattern defies explanations of luck or personnel or offensive system. Manning was the best in the league at avoiding sacks through coaching changes, team changes, rules changes, personnel changes, and every other variable outside of his control. The only constant is Manning himself.

The same applies to Dan Marino. Relative to his era, in which sack rates were higher, Marino was even more dominant than Manning with regard to avoiding sacks. He led all qualified passers every year from 1983-89, as well as 1994, 1997, and 1999. He was the best in the league at the beginning of his career, and the best at the end. Again, those teams have nothing in common but warm weather and Marino.

Here’s a different approach … the Buffalo Bills with Doug Flutie and Rob Johnson. From 1998-2000, Flutie had 1,063 pass attempts and 48 sacks (a 4.3% sack rate). Playing with the same teammates and coaches, Johnson had 447 attempts and 79 sacks (a 15.0% sack rate). The 2011 Denver Broncos featured both Kyle Orton (155 att, 9 sacks, 5.5%) and Tim Tebow (271 att, 33 sacks, 10.9%). Just last season, Arizona Cardinal QBs Drew Stanton (4.2%) and Blaine Gabbert (11.9%) had radically different sack rates. That’s not the offensive line. In 2007, Tom Brady got sacked on 3.5% of his dropbacks. In 2009, when Brady returned from injury, his sack rate was 2.8%. In between, Matt Cassel got sacked on 8.3% of his dropbacks for the same team. That’s not a group of blockers trying to get their backup QB killed, it’s a skill Brady had that Cassel didn’t.

Getting away from stats for a moment … recognizing that some sacks are the passer’s fault is fundamental to being a football fan. “Coverage sacks” are always the quarterback’s fault, and players like Manning and Marino and Namath had the awareness and skill to release the ball before getting hit even on a good pass rush or a missed block. They regularly turned sacks into pass attempts, saving yardage and often avoiding fumbles. My college team had a great quarterback, but everyone knew he took sacks instead of throwing the ball away, in a deliberate effort to keep his passer rating high. In the Greatest Show On Turf era, my friends and I — one of those friends being a Rams fan from St. Louis — were convinced that Kurt Warner did the same thing. Obviously some sacks are impossible to avoid — but most of them aren’t. QBs have primary control over their sack rates.

If you evaluate quarterbacks without considering sacks and fumbles, you’re not conducting a thorough evaluation. In extreme cases, like Marino and Namath, or Greg Landry and Neil Lomax (who have the highest sack rates in the top 100), an analysis that ignores sacks is deeply, irredeemably flawed.

Namath ranks here largely because of his excellence at avoiding sacks, and their cousin, fumbles. There are 11 quarterbacks among the Top 100 who average fewer than 1 fumble per 100 plays, where a play is defined as a pass attempt, rush attempt, or sack:

1. Bert Jones, 0.53
2. Johnny Lujack, 0.72
3. Peyton Manning, 0.74
4. Joe Namath, 0.82
5. Matt Ryan, 0.85
6. Joe Montana, 0.86
7. John Brodie, 0.97
8. Cam Newton, 0.98
9. Joe Theismann, 0.98
10. Ken Anderson, 0.99
11. Drew Brees, 0.99

Joe Namath was the best of his generation at avoiding sacks and fumbles, but he also had a great arm, and he was regarded as an excellent play-caller. He had devastating play-fakes, and against bump-and-run, he used to underthrow his receivers deliberately — what we now call the back-shoulder pass. Namath’s NY/A was by far the best of his generation, half a yard ahead of all his contemporaries. ^[9]Editor’s note: From ’65 to ’74, the first ten years of his career, Namath averaged 6.99 NY/A; Craig Morton was second at 6.48. He is one of the most efficient QBs in history at generating yardage. I know some people will never get past his high interception percentage, but they’re seeing a badly distorted picture of Namath’s career. He routinely faced extra defensive backs, he fumbled so seldom that his turnover rate was about average, he moved the chains like no one else in the game, and in his prime he was the best QB in pro football.

31. Kurt Warner — 12,202 — 20.5 — 4 – 32 – 128

The GSOT Rams have four Hall of Famers (Marshall Faulk, Orlando Pace, Aeneas Williams, and Warner), the most of any team since the Cowboys of the early ’90s. The Tony Dungy-era Colts will probably surpass them eventually, but the Rams — who won a single championship and only had three or four good seasons — have more HOFers than a lot of teams that were more successful. Furthermore, Isaac Bruce, London Fletcher, and Torry Holt — all of whom I would vote for — aren’t in.

The 2008-09 Cardinals could produce an outsized share of HOF inductions as well. Warner is in, Larry Fitzgerald is a lock, and Calais Campbell should get in. Anquan Boldin, Karlos Dansby, and Adrian Wilson are long-shots, but Warner played for two teams whose Hall of Fame reputations exceeded their results on the field.

30. Jim Hart — 16,076 — 21.16 — 8 – 36 – 288

Like Joe Namath, he ranks here partially because of his excellence at avoiding sacks. The top 10 passers of the 1970s, ranked by lowest sack percentage:

1. Jim Hart, 4.6
2. Fran Tarkenton, 6.5
3. Ken Stabler, 7.1
4. Dan Pastorini, 7.4
5. Terry Bradshaw, 7.7
6. Bob Griese, 8.6
7. Ken Anderson, 8.9
8. Roger Staubach, 9.37
9. Craig Morton, 9.43
10. Archie Manning, 9.6

Hart was comparably deficient in other areas — unlike Namath, his NY/A and fumble rates weren’t exceptional (though they were both significantly above average). Hart was sort of a high-level compiler. He had eight top-10-TSP seasons — the lowest-ranked player to do so — but only one top-3 season.

29. Jim Everett — 13,937 — 21.21 — 6 – 39 – 234

Relative to his statistics, Jim Everett was probably the most-snubbed quarterback ever in Pro Bowl voting. During Everett’s career (1986-97), 49 of 70 seasons (70%) with 1600 TSP resulted in Pro Bowl selections. Everett had five 1600-TSP seasons, including three 2000-TSP seasons, and made the Pro Bowl only once (20%).

I struggle to explain the lack of recognition for Everett. In each of his two 2000-TSP snub years, the Rams won double-digit games and made the playoffs. He played with good receivers like Henry Ellard and Flipper Anderson, true. But in 1989, when Everett led all QBs in TSP — and ranked first or second in the NFC in yards, TDs, and passer rating (for the second year in a row) — Joe Montana, Don Majkowski, Mark Rypien, and Randall Cunningham all made the Pro Bowl ahead of him. Cunningham, I admit, didn’t have much to work with. But Montana played with Jerry Rice and Roger Craig, Majkowski had Sterling Sharpe, and Rypien had maybe the greatest receiving trio of all time: the Posse, with Art Monk, Gary Clark, and Ricky Sanders.

Perhaps there was some shortcoming about Everett that doesn’t show up in the stats, but that voter insights could identify? The problem is that truly great quarterbacks, in their primes, are also among the worst snubs. Check out the list of 1800-TSP seasons that failed to garner Pro Bowl invitations. The list includes flukes like Steve DeBerg and Scott Mitchell, but also Hall of Famers like John Elway and Brett Favre, as well as lots of Jim Everett.

Jim Everett, 1988, 1989, 1994
Dan Marino, 1988, 1990
Steve DeBerg, 1990
Steve Young, 1991
Brett Favre, 1994
John Elway, 1995
Jeff George, 1995
Erik Kramer, 1995
Scott Mitchell, 1995

I suspect one of Everett’s problems ties back to a subject much discussed in this article. Everett got rid of the ball under pressure, throwing a high rate of interceptions, but with a low sack rate and not many fumbles. Among the top passers from 1985-94, only Dan Marino — who also made fewer Pro Bowls than one might predict based on his stats — had a lower sack rate than Everett. Instead of being praised for his smarts, though, Everett was criticized as not being tough.

28. Matt Ryan — 14,251 — 21.4 — 7 – 28 – 196

I doubt that many people with a historical understanding of professional football would name Matt Ryan among the 30 best quarterbacks of all time. I don’t, not yet.

Are contemporary players overrated by TSP? Possibly. In 2016, Brock Osweiler scored exactly zero TSP, and Adam Steele objected that Osweiler’s play was below replacement level. Some of that complaint was my fault for failing to clarify what I mean by “replacement level” — it’s the level of play you’d expect from an available free agent, not your top backup — and some of it was probably influenced by the narrative around Osweiler, who dramatically underperformed his generous contract. But that doesn’t mean that Adam was wrong. At a gut level, Osweiler’s performance felt pretty damn replaceable.

Similarly, three of the top four QBs in this ranking (Tom Brady, Drew Brees, and Peyton Manning) were continuously active from 2001-15. Fully a quarter of the top 28 were continuously active from 2008-15.

Here’s one way of measuring this. How many players in the TSP top 20, 40, 60, 80, and 100 were active in the following seasons?

1950: 3, 4, 5, 7, 8 (27)
1955: 3, 4, 6, 8, 9 (30)
1960: 5, 7, 11, 15, 17 (55)
1965: 4, 9, 14, 18, 20 (65)
1970: 5, 11, 16, 21, 23 (76)
1975: 4, 9, 14, 23, 28 (78)
1980: 3, 5, 9, 17, 26 (60)
1985: 7, 10, 13, 19, 28 (77)
1990: 5, 11, 14, 20, 24 (74)
1995: 5, 11, 16, 19, 23 (74)
2000: 4, 9, 16, 18, 22 (69)
2005: 5, 10, 19, 20, 25 (79)
2010: 5, 10, 13, 15, 18 (61)
2015: 4, 8, 10, 14, 16 (52)

The highest overall are 2005 (79), 1975 (78), 1985 (77), 1970 (76), and 1990 and ’95 (both 74). While this method is imprecise, it seems to me that TSP treats this era about the same as every other era.

27. Bobby Layne — 15,510 — 21.6 — 9 – 51 – 459

Underrated in 2018. A two-time NFL champion and the best dual-threat passer-runner of his generation, Layne was handicapped throughout his career by an offensive supporting cast that didn’t nearly measure up to those of contemporary QBs like Otto Graham, Y.A. Tittle, John Unitas, or Norm Van Brocklin. Layne was a first-ballot Hall of Famer in 1967, going in ahead of contemporaries like Van Brocklin and Tittle (both 1971, and neither first-ballot). Like many of the other QBs featured in this article, Layne is underrated by analyses that focus on traditional passing stats. He didn’t take a lot of sacks, and he led all QBs in rushing three times. ^[10]That’s tied (with Steve Young and Bobby Douglass) for 7th-most in the Modern Era. The leaders are Randall Cunningham, Tobin Rote, Michael Vick, Fran Tarkenton, Cam Newton, and Steve  … Continue reading He ranks 19th all-time by Top-10-Points, and 15th all-time by Year-Points, ratings which I think are more reflective of his greatness than 27th by Career Value.

26. John Hadl — 15,238 — 21.730 — 8 – 35 – 280

In this ranking, John Hadl is the highest-rated player with AFL experience. The full list:

26. John Hadl, 21.7
32. Joe Namath, 19.6
38. Daryle Lamonica, 17.7
40. Len Dawson, 17.6
63. Tobin Rote, 13.4
69. Bob Griese, 12.7
80. George Blanda, 11.2

The TSP formula treats the later AFL as a substantially stronger league than the early AFL, which produces rankings that are counter-intuitive but, I believe, well-founded.

25. Jim Kelly — 15,425 — 21.731 — 8 – 34 – 272

Jim Kelly began his pro career in the USFL, playing two seasons with the Houston Gamblers. Kelly passed for 83 TDs with the Gamblers, which is impressive even in 18-game seasons, and he started immediately for the Bills when the rival league collapsed. Kelly’s 3,593 passing yards in 1986 set an NFL rookie record, and he made the Pro Bowl the following season. Kelly was the best quarterback in the USFL — a league which included Steve Young, Doug Flutie, Brian Sipe, Greg Landry, Doug Williams, and Bobby Hebert — and he validated that performance with immediate success in the established league. When we evaluate Kelly, it’s important to remember that he had two good years which don’t show up in the official stats: he was obviously a good pro QB in 1984-85, he just wasn’t in the NFL.

24. Roman Gabriel — 15,982 — 22.3 — 5 – 33 – 165

I believe that any time I have ever ranked the greatest quarterbacks of all time, for publication or my own amusement, I have rated Roman Gabriel and John Hadl back-to-back. They were exact contemporaries, born less than six months apart, and both played from 1962-77. They were even traded for each other, more or less, in 1973. But they’re surprisingly hard to compare, because they began in different leagues and they had substantially different styles and strengths.

Let’s start with raw stats. Sacks and sack yardage count toward the attempt and yardage totals.

Player   Att    Yds     NY/A  TD   INT  Rating  RshYd   RshTD Fmbl
Gabriel  4,856  26,733  5.51  201  149  74.3    1,304   30    105
Hadl     4,971  30,935  6.22  244  268  67.4    1,112   16     73

Hadl produced many more yards and TDs, with a far better net average. Gabriel had many fewer turnovers, despite his higher fumble frequency. Gabriel was a cautious, high-percentage passer, a George Allen product. Hadl was a gunslinger, with the same mentality as his star WR, Lance Alworth. Gabriel had the strongest arm of his generation; Hadl was a great touch passer. For contemporary players with similar-length careers and roughly the same number of pass attempts, they’re surprisingly difficult to compare.

Career Value is my favorite statistical measure of QBs, and Gabriel rates slightly ahead by that metric, as well as by Adjusted TSP. By Year-Points, though, Hadl scores dramatically higher. Huh? These guys were competing against each other for Year-Points; where does this disparity come from? Hadl had more good seasons, but Gabriel’s best seasons were better than Hadl’s. The chart below shows how many seasons each player had with 1000 TSP, 1500 TSP, and 2000 TSP.

Gabriel: 6, 4, 2
Hadl: 9, 5, 0

Thus, Hadl had eight top-10 seasons, compared to just five for Gabriel. But because Gabriel’s best seasons were so excellent, their Top-10-Points are essentially equal, 35-33. And because TSP emphasizes peak production through the ^1.85 exponent, Gabriel’s big seasons — he was over 1900 TSP three times — count for more than Hadl’s consistency. Which was actually better? I still don’t know. If I had to pick, today I’d say Gabriel. I’ve gone the other direction in the past.

23. Ben Roethlisberger — 17,593 — 22.8 — 7 – 26 – 182

Several times in this series, I’ve referenced the “Bob Griese Problem.” It’s my belief that high-efficiency, low-volume passers are underrated by TSP. This rating system is designed to value both production and efficiency. Obviously a player who is inefficient should not score well, because he does not put his team in position to win. Similarly, a player with limited production is not making significant contributions to his team. He is a game manager, a term we all recognize in reference to a modestly talented passer who takes few chances, hoping to ride a ground game and defense into the postseason. Neil O’Donnell, perhaps the quintessential game manager, for years held the all-time record for lowest interception percentage, but he ranks 110th in Career Value. That’s a credit to the TSP formula, I believe.

However, this principle causes TSP to punish great players whose stats are constrained by the run-based nature of their offenses. Griese, a Hall of Famer who ranks 69th by CV, is the best example. He was a two-time first-team All-Pro who rarely passed on 1st and 2nd downs because he played on one of the best teams of all time, the Miami Dolphins of the early ’70s. Those teams had a great defense that kept scores low, a legendary backfield, and probably the best interior offensive line of all time.

Statistically, however, it is Ben Roethlisberger who most clearly demonstrates that the Bob Griese Problem is, in fact, a problem. Big Ben began his career with Bill Cowher, on teams that won with defense and played offense to control the ground and the clock. In each of his first two seasons, Roethlisberger had fewer than 300 pass attempts. He didn’t throw 500 times in a season until 2009, his sixth year in the league. He didn’t rank among the top 10 in attempts until 2013.

That limited Roethlisberger’s TSP. His first season, when he broke the rookie record for passer rating (98.1) and went 13-0 as a starter in the regular season, Roethlisberger scored 1158 TSP (1.3 CV). The next season, with a 98.6 rating and a Super Bowl victory, he improved to 1288 TSP (1.6 CV). In 2007, Ben’s 104.1 rating was second only to MVP Tom Brady, but with only 404 attempts (and 47 sacks), he registered just 1317 TSP (1.7 CV).

Obviously Roethlisberger was better than those scores implied, but statistically, he looked like a part-time player. In ’04 and ’05, you could be forgiven for imagining that he only played half the season. His early stat lines have a lot in common with Bart Starr’s in the ’60s.

When Ben finally became a high-volume passer — which roughly lines up with when Todd Haley became his offensive coordinator, Le’Veon Bell entered the league, and Antonio Brown turned into Steve Largent — his efficiency remained high while his production exploded, and he started making the Pro Bowl every year. I believe that if Griese and Starr and Troy Aikman had played for Todd Haley, they would have production and efficiency commensurate with their talents.

This argument can also be illustrated with the Steelers’ Hall of Fame receivers of the ’70s and ’80s, John Stallworth and Lynn Swann. On the surface, Stallworth’s statistics are vastly superior to Swann’s. Stallworth had three 1,000-yard seasons, including 1,395 in 1984. Swann never had a 900-yard receiving season.

Those numbers are somewhat misleading. Although both players were rookies in 1974, Swann had his best seasons early — when the season was 14 games, the rules favored defense, and Pittsburgh was a rushing team, with HOF running back Franco Harris in his prime. Stallworth, who for many years had trouble staying healthy, had his best seasons when the season was 16 games, the rules favored passing, and Harris was no longer the focal point of the offense.

Player statistics are powerfully dictated by the circumstances under which those stats are produced, including league environment and season length (which TSP controls for) as well as coaching philosophy and quality of teammates (which TSP can’t discern). In Roethlisberger’s first seven seasons, which saw the Steelers play in three Super Bowls, he accrued 9.3 CV. In the seven seasons since, when the Steelers reached no Super Bowls, Big Ben has added 13.5 CV. Has Ben been 45% better, or 45% more valuable, since his team stopped playing in championships? Of course not. It’s the Bob Griese Problem.

22. Boomer Esiason — 15,881 — 23.9 — 6 – 41 – 246

Jim Kelly and Boomer Esiason were contemporaries. Kelly played in the NFL from 1986-96, Esiason from 1984-97. Statistically, they’re similar. As usual, the chart below includes sacks in the attempt and yardage columns.

Player   Att    Yds     NY/A  TD   INT  Rating RshYd RshTD Fmbl
Esiason  5,523  35,482  6.42  247  184  81.1   1,598   7   123
Kelly    5,102  33,040  6.48  237  175  84.4   1,049   7    76

Esiason has slightly higher Adjusted TSP, Career Value, and Top-Ten-Points. They both made four Pro Bowls, and both were first-team All-Pro once. The obvious difference is that Kelly played in four Super Bowls, to Esiason’s one. Only slightly less obvious is that Kelly played with a Hall of Fame coach, a Hall of Fame running back, two Hall of Fame wide receivers, and the best offensive line in the AFC. Esiason’s leading receivers were Eddie Brown, Tim McGee, and Rodney Holman.

In a 1991 interview with Paul Zimmerman, Hall of Fame coach Sid Gillman, the godfather of the modern passing game, discussed his favorite contemporary QBs:

There are his absolute blue-chippers, “guys who can sustain an offense for you”: John Elway, Joe Montana, Randall Cunningham, Warren Moon (“a beautiful passer; I don’t like the run-and-shoot, but with that guy running it, it’s so pretty to watch”), Jim Kelly and Dan Marino, who drives Sid nuts. “Totally undisciplined,” Gillman says of the Miami Dolphin passer, “but, my god, what a talent: what a quick drop, quick hands. But when he takes his drop, he’s all over the place.”

And then there’s his favorite, Boomer Esiason. “Technically the most perfect,” Gillman says. “He does all the things I tried to get my quarterbacks to do … I just love watching the guy play.”

Esiason isn’t in the Hall of Fame because he only had six good seasons, in an era when contemporaries like Elway, Marino, and Moon sustained success for twice as long, but at his best he was a truly great quarterback. Combined TSP, six best seasons:

Dan Marino — 15,373
Joe Montana — 13,940
John Elway — 11,680
Warren Moon — 11,349
Boomer Esiason — 11,321
Randall Cunningham — 10,127
Jim Kelly — 9,884

Boomer rates essentially even with Elway and Moon, and comfortably ahead of Kelly. The difference is that Esiason had only six seasons with 1100 TSP, while Marino (15), Montana (11), Elway (12), Moon (10), and Kelly (10) all had substantially more. Kurt Warner notwithstanding, the HOF voters value consistency and longevity, and Boomer was more of a shooting star. I wonder whether his HOF candidacy would have more momentum if the chronology of his career were re-arranged a little, to block his comeback seasons in 1993 and 1997 together with his run of excellence from 1985-89, forming a longer peak in which those seasons seem less like aberrations and more like a cohesive pattern of high-level play.

21. Philip Rivers — 15,819 — 24.2 — 8 – 52 – 416

Philip Rivers, Ben Roethlisberger, and Eli Manning are linked in the public’s mind, as the three quarterbacks chosen near the top of the 2004 NFL Draft. I wrote a lengthy article about the trio four years ago. Some of the specifics are outdated now, but the conclusions hold up with very little adjustment. Stats through 2017:

Player         Att    Yds     NY/A  TD   INT  Rating RshYd  RshTD Fmbl
Manning        7,755  49,200  6.35  339  228  83.5     540    6   115
Rivers         6,872  48,350  6.99  342  166  94.8     573    3    99
Roethlisberger 6,970  47,918  6.84  329  174  94.0   1,252   16    91

Manning has the most net yardage, but that’s unlikely to last, since Rivers and Roethlisberger are still playing well and Eli appears to be on his last legs. Manning also has the worst NY/A, rating, TD/INT +/-, and fumble total, all by wide margins. He has the lowest TSP and Career Value, the fewest Top-10 seasons and Top-10-Points, and the fewest Pro Bowls. He even has the worst record: Manning is now 111-103 as a starter (.519), compared to 135-63 for Roethlisberger (.681) and 106-86 for Rivers (.552). Manning’s only argument is that he, not the Giants’ defense, was the critical factor in New York’s postseason success, and that doesn’t hold up. Giants fans have been reduced to arguing that Kevin Gilbride was the worst offensive coordinator in the history of organized football, and single-handedly responsible for the illusion of Manning’s inferiority relative to his ’04 classmates. That’s a pretty barren argument, too, after savior Ben McAdoo was unable to revive Eli’s career.

What Eli Manning fans should really hope over the next couple of years is for Rivers to solidify his Hall of Fame case. Roethlisberger is probably a lock at this point, and if he and Rivers go to Canton, Manning will eventually complete the triangle. There are so many Giants fans, and so many people blinded by team accomplishments, that there will be riots when Rivers gets enshrined before Eli; I can’t envision a scenario where Rivers gets voted in and Eli doesn’t. Ben will reach Canton first, but Rivers and Manning will go in or not — and I’d guess yea — as a pair, probably within three years of one another.

My experience is that if you ask a football fan who’s not into analytics to rank the Class of ’04 QBs, you’ll get Ben and Eli pretty even at the top, with Rivers a distant third. That should be astonishing, but our sports culture is incredibly backward in how it evaluates QBs.

Next week, we’ll evaluate the 20 greatest statistical QBs of pro football’s Modern Era.