Tuesday, July 10, 2018
Best Statistical QBs: HOF Data
This is the penultimate article in a seven-part series. It is a supplement to my 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. Over the last five weeks, I've written about the top 100 quarterbacks of the Modern Era, as ranked by TSP:
This article presents a different way of evaluating quarterbacks. It is data-heavy, so if you're just here for my prose, I'm sorry to disappoint. The chart below shows those same 100 QBs, along with each player's: Adjusted TSP, Career Value, Seasons among the top 10 in QB-TSP, Top-10-Points, Year-Points, 500-TSP seasons, 1,000-TSP seasons, 1,500-TSP seasons, 2,000-TSP seasons, 2,250-TSP seasons, 2,500-TSP seasons, Pro Bowls, All-Pro honors, and championship wins and appearances. For those last two categories, I differentiated between First-Team All-Pro selections (usually 3 points) and Second-Team All-Pro selections (usually 1 point), and between Super Bowl wins (usually 3 points) and losses (usually 1 point). There's a little more to it than that, but it doesn't make for very interesting reading.
This isn't just a data dump, though. My motivation in compiling these numbers was to determine what they signify relative to the Pro Football Hall of Fame. For each statistic, I'll present the following findings:
1. Above what point do 50% of quarterbacks make the Hall of Fame?
2. Above what point do 75% of quarterbacks make the Hall of Fame?
3. Above what point do 90% of quarterbacks make the Hall of Fame?
4. Who is the median HOF QB and what is his score in this category?
5. Who is the best of all-time in this category?
6. Among HOFers, who is the lowest scorer in this category?
I counted the following players, some of whom are not in the Hall of Fame yet, as HOF QBs: Troy Aikman, Terry Bradshaw, Brett Favre, Otto Graham, Bob Griese, Len Dawson, John Elway, Dan Fouts, Sonny Jurgensen, Jim Kelly, Bobby Layne, Dan Marino, Joe Montana, Warren Moon, Joe Namath, Ken Stabler, Bart Starr, Roger Staubach, Fran Tarkenton, Y.A. Tittle, Johnny Unitas, Norm Van Brocklin, Kurt Warner, Steve Young, Tom Brady, Drew Brees, Peyton Manning, and Aaron Rodgers.
That excludes: Pre-Modern QBs like Sammy Baugh and Sid Luckman; Bob Waterfield, who really should fall into the same category; and George Blanda, who is in Canton more for his kicking than his passing. I judged Brady, Brees, Manning, and Rodgers to be HOF locks, but I deliberately omitted potential HOFers like Ben Roethlisberger, Philip Rivers, Eli Manning, and Matt Ryan. Along with those four, I evaluated several possible HOF snubs — Ken Anderson, Randall Cunningham, and Donovan McNabb — to see how they match up in each statistical category. It's not an entirely fair comparison, since so much of the argument for Cunningham and McNabb derives from their subpar offensive teammates and the premise that their statistics underrepresent their excellence. I'm trusting you, readers, to put all these figures in context. Football is a complicated sport, and its stats cry out to be analyzed, not accepted at face value.
Roethlisberger, Rivers, Manning, and Ryan are included in the chart below, but they are excluded from the calculations that follow; they don't count as Hall of Famers or non-Hall of Famers, since their HOF futures remain unclear. Other active players, like Russell Wilson, may be headed for Canton, but clearly haven't done enough yet to merit serious consideration. They count as non-HOFers.
If you're still interested, here's the big chart. It's organized by Career Value.
Adjusted TSP
50% mark: 11,069 (Joe Theismann)
75% mark: 14,225 (Len Dawson)
90% mark: 18,157
Median HOF: 19,263
Highest: 34,057 (Peyton Manning)
Lowest HOF: 11,234 (Ken Stabler)
To make sure we're clear... Joe Theismann has 11,069 TSP. Including Theismann, half of the players with QB-TSP that high are in the Hall of Fame. That's not terribly meaningful, though, because most of them cluster near the top. Reaching 11,000 TSP does not imply a 50% chance of Hall of Fame election. Somewhat more meaningful is 14,225 career TSP; 75% of players who reach that level are in Canton. The 90% figure falls between Y.A. Tittle (19,087) and John Brodie (17,839). None of those percentiles are calculated including Philip Rivers, Ben Roethlisberger, and Matt Ryan, who are above the first two standards, or Eli Manning, who meets the 50% standard but not 75%. Those four players are excluded from all the categories below, as well. None of the quartet has reached the 90% threshold, though Rivers and Roethlisberger are both likely to surpass it in 2018.
Among the 28 HOF QBs I evaluated, Warren Moon ranks 14th in this statistic and Norm Van Brocklin ranks 15th. The average between their scores comes to 19,263, a figure which roughly represents a "normal" Hall of Famer's production in this metric. Peyton Manning predictably has the highest score of any QB, while Ken Stabler, who only had four or five good seasons, ranks the lowest among HOFers.
Prospective HOF: Anderson [20,133], Rivers [17,668], Roethlisberger [17,593], Ryan [14,251], McNabb [13,717], Cunningham [12,557], Manning [12,113]
If Adjusted TSP were your preferred HOF criterion, you'd expect Ken Anderson to get in with little difficulty. Rivers and Roethlisberger would have a good chance, with Ryan and McNabb likely a little short. Cunningham and Manning, as judged by this stat, do not have strong HOF cases. Throughout the "Prospective HOF" sections, "Manning" refers to Eli.
Career Value
50% mark: 14.3 (Billy Wade)
75% mark: 20.5 (Kurt Warner)
90% mark: 27.9 (John Brodie)
Median HOF: 31.7
Highest: 67.7 (Peyton Manning)
Lowest HOF: 12.7 (Bob Griese)
I've written extensively on this metric, including an organized ranking over the last five weeks, so I'm not going to spend a lot of space on this stat. If you're interested, there's already a lot out there. The median HOF mark falls between Norm Van Brocklin (31.9) and Y.A. Tittle (31.5).
Prospective HOF: Anderson [34.3], Rivers [27.3], Roethlisberger [22.8], Ryan [21.4], Cunningham [18.8], McNabb [18.2], Manning [14.2]
Judging by CV, which is my favorite stat for assessing QB careers, Anderson should be a cinch Hall of Famer. Rivers, Roethlisberger, and Ryan have decent arguments already — especially Rivers — while Cunningham and McNabb are long shots, with Eli Manning not even meriting discussion. I'll remind you, though, that all of these statistics are produced in unique contexts. While TSP (and therefore CV) is era-adjusted, it doesn't know anything about coaches, blockers, receivers, or other factors that can affect a QB's statistical output. Furthermore, TSP is a regular-season metric.
Top 10 Seasons
50% mark: 4
75% mark: 6
90% mark: 8
Median HOF: 9
Highest: 15 (Peyton Manning)
Lowest HOF: 3 (Bob Griese)
PFHOF voters really value counting stats and longevity. Allowing for isolated off years or seasons lost to injury, Hall of Fame QBs normally remain among the best at their position for over a decade. It's a tough standard.
Prospective HOF: Anderson [8], Rivers [8], Roethlisberger [7], Ryan [7], Manning [6], Cunningham [5], McNabb [5]
If all you looked at was the number of seasons among QB-TSP's top 10, Anderson and Rivers would be borderline HOFers, Roethlisberger and Ryan would be long shots, and the other three would be really long shots.
Top-10-Points
50% mark: 21
75% mark: 33 (Roman Gabriel)
90% mark: 41 (Boomer Esiason)
Median HOF: 59
Highest: 124 (Peyton Manning)
Lowest HOF: 15 (Bob Griese)
There is a giant gap between 20th and 21st. The top 20 all have at least 50 points, and 19 of those 20 are in the Hall of Fame or not yet eligible. Esiason ranks 21st, with 41 points. It's a pretty clear delineation of the Hall of Fame standard.
Prospective HOF: Anderson [57], Rivers [52], Ryan [28], Cunningham [26], Roethlisberger [26], McNabb [22], Manning [11]
Roethlisberger rates noticeably worse by this measure. He has four seasons ranked between 8-10, so even though his number of Top 10 seasons is consistent with his career production, his Top-10-Points come in low. Big Ben also has quite a few near misses: 11, 12, 12, 13, and 13, illustrating the problem with base-10 cutoffs. This is why TSP and CV are better measures of a player's career. Top-10-Points can help identify players potentially overrated or underrated by those metrics, but it's a supplement to those stats; it should never replace them.
Eli Manning provides an even more striking illustration than Roethlisberger of this system's vulnerabilities. He has six top-10 seasons, which is good, but his individual ranks were: 10, 10, 5, 10, 10, 10. He has no near-misses, no seasons ranked between 11-13. If I used a cutoff of 8, or 9, or 12, Manning's rank in these metrics would drop considerably.
Year-Points
50% mark: 85 (Phil Simms)
75% mark: 165 (Roman Gabriel)
90% mark: 279
Median HOF: 473.5
Highest: 1,860 (Peyton Manning)
Lowest HOF: 45 (Bob Griese)
There are five players with at least 1,079 Year-Points, and no one else with more than 902. The top five are Peyton Manning, Dan Marino (1,274), Drew Brees (1,248), Fran Tarkenton (1,092), and Tom Brady (1,079). Johnny Unitas is sixth.
Prospective HOF: Anderson [456], Rivers [416], Ryan [196], Roethlisberger [182], Cunningham [130], McNabb [110], Manning [66]
By this measure, Anderson and Rivers rate as Hall of Fame-quality. None of the others come particularly close, though Ryan and Roethlisberger could climb quickly.
500-TSP Seasons
50% mark: 8
75% mark: 13
90% mark: 13
Median HOF: 11
Highest: 18 (Fran Tarkenton)
Lowest HOF: 7 (Kurt Warner)
500 TSP is such a low standard that this not a meaningful measure with regard to true greatness or Hall of Fame credentials. A 500-TSP season is the minimum standard for a productive season, basically just a starter who isn't in the bottom 10% of the league. The HOF floor (Warner's 7, or maybe Namath, Stabler, and Staubach's 8, if you regard Warner as a fluke) is the most interesting figure here.
Prospective HOF: Roethlisberger [14], Anderson [11], Rivers [11], Manning [10], McNabb [10], Ryan [10], Cunningham [7]
Cunningham's low number of productive seasons may explain why his HOF candidacy has attracted so little interest. The other six are all within a normal range for Hall of Famers.
1,000-TSP Seasons
50% mark: 5
75% mark: 8
90% mark: 9
Median HOF: 9
Highest: 16 (Peyton Manning)
Lowest HOF: 5 (Ken Stabler)
There's a pretty significant correlation between this category and Top 10 Seasons. The latter is a slightly higher standard, but the 10th-ranked season almost always falls between 1,000 and 1,500 TSP. Since 1962, the only exceptions were 1995 (1,634 TSP) and 2006 (862 TSP).
Prospective HOF: Roethlisberger [12], Rivers [10], Anderson [9], McNabb [9], Ryan [8], Cunningham [7], Manning [7]
Roethlisberger really shines by this measure. He's had a lot of statistically above-average seasons; he's one of the most reliably productive passers in history.
1,500-TSP Seasons
50% mark: 3
75% mark: 4
90% mark: 5
Median HOF: 7
Highest: 15 (Peyton Manning)
Lowest HOF: 1 (Bob Griese)
Not counting 2011, when he missed the entire season with a neck injury, Peyton Manning played 17 seasons in the NFL. The only seasons in which he did not score at least 1,500 TSP were his first (1998) and his last (2015).
Prospective HOF: Anderson [5], Rivers [5], Ryan [4], Roethlisberger [3], Cunningham [3], McNabb [2], Manning [1]
This might be the TSP stat that most closely correlates with Hall of Fame status. Eighteen of the 19 players with 6 seasons at this level are in the Hall of Fame (Boomer Esiason is the only exception). What impresses the Hall of Fame voters most is having half a dozen or more really good seasons. That's good news for Philip Rivers.
2,000-TSP Seasons
50% mark: 1
75% mark: 2
90% mark: 3
Median HOF: 3
Highest: 10 (Peyton Manning)
Lowest HOF: 0
Fifty-six quarterbacks have had a 2,000-TSP season. Twenty of them are in the Hall of Fame (including George Blanda), eight are not yet eligible (the Big Four plus Rivers, Roethlisberger, Ryan, and Carson Palmer), and 28 have been passed over by the voters. But essentially, if you ever have a 2,000-TSP season, there's about a 50% chance you'll go on to the Hall of Fame.
Prospective HOF: Anderson [4], Rivers [3], Cunningham [2], McNabb [1], Roethlisberger [1], Ryan [1], Manning [0]
I get annoyed when people suggest that Randall Cunningham was a one-year wonder in 1998. In 1990, he was top-5 in pass TDs, TD/INT +/-, and passer rating, while rushing for 942 yards, an 8.0 average, and 5 TDs. Pro Football Writers of America and Pro Football Weekly named Cunningham First-Team All-Pro, and the former named him MVP as well. United Press, which named All-Conference teams rather than All-Pro, chose Cunningham as its All-NFC quarterback, ahead of AP MVP Joe Montana. Beyond that, ancedotally, did you ever watch Randall Cunningham play between 1988-92? He was a unique talent.
If you're keeping track of the "prospective" group, Eli Manning has sole possession of last place in six of the nine categories so far.
2,250-TSP Seasons
50% mark: 1
75% mark: 1
90% mark: 2
Median HOF: 1.5
Highest: 5 (Peyton Manning and Joe Montana)
Lowest HOF: 0
Forty-one modern QBs have combined for 77 seasons at this level. Seventeen have reached it more than once. Ken Anderson, John Brodie, and Daunte Culpepper are the only non-HOFers to do so.
Prospective HOF: Anderson [4], Cunningham [1], Ryan [1], Rivers [0], Roethlisberger [0], McNabb [0], Manning [0]
It's pretty remarkable that Kenny Anderson isn't in the Hall of Fame. The other QBs with four 2,250-TSP Seasons are Peyton Manning, Joe Montana, Drew Brees, Otto Graham, Dan Marino, John Unitas, and Steve Young. Those are inner-circle Hall of Famers, not borderline at all.
2,500-TSP Seasons
50% mark: 1
75% mark: 1
90% mark: 1
Median HOF: 1
Highest: 4 (Otto Graham and Steve Young)
Lowest HOF: 0
Twenty-seven modern QBs have combined for 42 seasons at this level. Seventeen of those QBs are in the Hall of Fame or locks once they're eligible, and an 18th (Matt Ryan) is too early in his career to make a confident projection one way or the other. The remaining nine, in chronological order, are: Johnny Lujack (1949), Milt Plum (1960), John Brodie (1970), Ken Anderson (1975, 1981, 1982), Bert Jones (1976), Brian Sipe (1980), Randall Cunningham (1998), Jeff Garcia (2000), and Daunte Culpepper (2004).
Prospective HOF: Anderson [3], Cunningham [1], Ryan [1], Rivers [0], Roethlisberger [0], McNabb [0], Manning [0]
Matt Ryan's 2016 season was a really nice foundation for a Hall of Fame career. I'm sure the voters wouldn't take him seriously yet, but another 2,000-TSP season would make 2016 look less like an outlier, significantly boosting his credibility with the panel.
Pro Bowls
50% mark: 3
75% mark: 4
90% mark: 6
Median HOF: 7
Highest: 14 (Peyton Manning)
Lowest HOF: 3 (Terry Bradshaw)
This category is unfair to QBs who played before 1950, since there were no Pro Bowls held in the late '40s. Most notably, Otto Graham would very likely have nine All-Star appearances rather than five.
Since Graham is the only Modern-Era HOFer affected, though, this category correlates very strongly with Hall of Fame induction. Of the 17 players with 7 or more Pro Bowl (or AFL All-Star Game) appearances, 16 are in Canton or not yet eligible; the only exception is the AFL's Jack Kemp. If you lower the threshold to six Pro Bowls, 22 of 25 eligible players are in; the exceptions are Kemp, John Hadl (also AFL), and Donovan McNabb. As you can probably tell, AFL All-Star appearances are overrated here; I probably should have counted them as one-half instead of full value.
Prospective HOF: Rivers [7], McNabb [6], Roethlisberger [6], Anderson [4], Cunningham [4], Manning [4], Ryan [4]
This tells you a lot about the context in which players like McNabb and Anderson produced their stats, or at least how that context was perceived at the time. It's incredible how quickly we've forgotten that McNabb was arguably the 2nd-best QB of the early 2000s.
All-Pro Points
50% mark: 1
75% mark: 5
90% mark: 7
Median HOF: 8
Highest: 29 (Peyton Manning)
Lowest HOF: 0 (Roger Staubach)
If these numbers look wacky, please remember that I'm not just counting All-Pro selections! In most years, a first-team All-Pro selection and an MVP is worth four points toward this tally.
Prospective HOF: Cunningham [7], Anderson [5], Ryan [4], Manning [0], McNabb [0], Rivers [0], Roethlisberger [0]
Cunningham was named Second-Team All-Pro in 1988 (+1), First-Team All-Pro (non-AP) and MVP in 1990 (+2), Second-Team All-Pro in 1992 (+1), and First-Team All-Pro in 1998 (+3).
Championship Points
50% mark: 1
75% mark: 4
90% mark: 5
Median HOF: 5.5
Highest: 20 (Otto Graham)
Lowest HOF: 0
As with the All-Pro category, players can earn points in seasons their team did not win a championship, and can earn up to 3 points for years in which their teams did. Dan Fouts, Sonny Jurgensen, and Warren Moon are the Hall of Famers who score zero in this category. That's particularly unfair for Moon, who won five professional championships in Canada. I might also argue for Donovan McNabb and Ken Stabler, both of whom reached five conference championship games but only one Super Bowl, to rate a little higher, since they consistently had their teams in the championship mix.
Prospective HOF: Roethlisberger [7], Manning [6], Anderson [1], McNabb [1], Ryan [1], Cunningham [0], Rivers [0]
Eli Manning's Hall of Fame case rests almost entirely on this category, which is really a team achievement rather than an individual one. Manning's teams have never won a playoff game in which they allowed more than 20 points.
Hall of Fame Index
Okay, new stat here. This is designed to approximate popular opinion and HOF voting standards. I believe it has predictive value as a gauge of conventional wisdom and HOF probability, but I beg you not to treat it as a meaningful measure of skill or accomplishment. The formula is annoying to calculate without a spreadsheet:
(Adjusted TSP / 3000) + (Career Value / 1.8) + (3 * Pro Bowls) + (1.75 * All-Pro Points) + (4 * Championship Points) + (1.5 * Super Bowl era Conference Championship losses) + [ (Rookie Year + Final Year - 4000) / 25 ]
As an example, Matt Schaub has 7,549 TSP, 9.52 CV, 2 Pro Bowls, no All-Pro selections or Conference Championship Game appearances, and has played from 2004-17. His score is 2.52 + 5.29 + 6 + 0 + 0 + 0 + 0.84 = 14.65.
The results are not perfect; there are some hiccups. However, the formula acknowledges the impact of stats, postseason honors, team success, and recency bias. To more accurately reflect modern perceptions, I halved the value of all Championship Points prior to the Super Bowl era and of all AFL All-Star Games.
For this one, I'll just give you the list. I rounded to whole numbers, but there are a lot of decimals hidden; there are no ties.
1. Tom Brady, 182
2. Peyton Manning, 176
3. Joe Montana, 131
4. Otto Graham, 130
5. Johnny Unitas, 124
6. Brett Favre, 112
7. Dan Marino, 99
8. John Elway, 99
9. Drew Brees, 98
10. Steve Young, 90
11. Fran Tarkenton, 83
12. Terry Bradshaw, 79
13. Bart Starr, 77
14. Aaron Rodgers, 73
15. Roger Staubach, 72
16. Norm Van Brocklin, 72
17. Troy Aikman, 71
18. Bob Griese, 70
19. Y.A. Tittle, 68
20. Ben Roethlisberger, 68
21. Dan Fouts, 66
22. Kurt Warner, 62
23. Bobby Layne, 61
24. Jim Kelly, 56
25. Warren Moon, 52
26. Ken Stabler, 51
27. Joe Namath, 50
28. Len Dawson, 49
29. Ken Anderson, 49
30. Eli Manning, 49
31. Sonny Jurgensen, 46
32. Philip Rivers, 44
33. Rich Gannon, 43
34. Donovan McNabb, 43
35. Matt Ryan, 42
36. Boomer Esiason, 41
37. Joe Theismann, 40
38. Randall Cunningham, 40
39. Russell Wilson, 39
40. Daryle Lamonica, 39
41. John Brodie, 36
42. John Hadl, 36
43. Roman Gabriel, 33
44. Phil Simms, 32
45. Cam Newton, 31
46. Jim Plunkett, 30
47. Steve McNair, 30
48. Jack Kemp, 29
49. Tony Romo, 29
50. Jim Hart, 29
Other notable: Drew Bledsoe, 28; Charlie Conerly, 28; Carson Palmer, 27; Earl Morrall, 26; George Blanda, 24; Daunte Culpepper, 22; Vinny Testaverde, 21; Jim Everett, 20; Alex Smith, 19; Lynn Dickey, 5.
To the extent this stat is meaningful, it suggests that Roethlisberger is a PFHOF lock, Anderson and Eli are borderline, and Rivers needs another Pro Bowl season. Ryan isn't there yet but probably will be in a few years, while McNabb and Cunningham shouldn't get their hopes up.
I cannot overemphasize that I think this statistic, while interesting as an approximation of conventional wisdom, is worse than useless as a tool for meaningful analysis. If you use it to measure HOF worthiness, you will reach deeply flawed conclusions. It is specifically and exclusively designed to gauge the worst kind of conventional wisdom; the target audience is Skip Bayless.
There is one article remaining in this series: an update of my personal top 100 QBs of the Modern Era. Please check back next week.