Ranking the Best Statistical QBs of 2017

This is the third time in the last four years that I've written about my preferred stat for evaluating NFL quarterbacks, QB-TSP. In this post, you'll find scores from the 2017 season, plus another way of using TSP.

If you're not already familiar with the stat, I'd encourage you to read about how it works, but if you're in a hurry, it is a purely statistical ranking, not my opinion. TSP measures production above replacement level, with "replacement level" defined as the level of play you'd expect from an available free agent (not your top backup).

A good example last season was Jay Cutler, lured from retirement to play for Miami after Ryan Tannehill got hurt. Robert Griffin and Johnny Manziel didn't play last season, but either one would have been a replacement player, as would an undrafted college senior. Anyone who you're not sure whether or not they were still on a roster, like Kellen Clemens or Kellen Moore, is probably right around replacement level.

Here are rough explanations of single-season TSP and how it translates to Career Value:

* Zero TSP (0 CV) indicates replacement-level performance, on the fringe of being playable. 2017 example: Trevor Siemian.

* 500 TSP (0.3 CV) is an inconsequential season, an ineffective starter or a good part-time player. 2017 examples: Jacoby Brissett, Aaron Rodgers.

* 1000 TSP (1 CV) is an average season. The player had some value to his team, but he wasn't a Pro Bowl-quality performer. 2017 examples: Blake Bortles, Dak Prescott.

* 1500 TSP (2 CV) is a good season, a top-10 season, a borderline Pro Bowl season. This is a positive contribution to any player's résumé. 2017 examples: Ben Roethlisberger, Matthew Stafford.

* 2000 TSP (3.5 CV) is a great season. It's a top-5 performance, the player almost always makes the Pro Bowl, and he'll usually generate some all-pro support. 2017 examples: Alex Smith, Tom Brady.

* 2500 TSP (5.5 CV) is an exceptional season. These only occur about twice every three years. Most of them were first-team All-Pro, and about half were named league MVP. 2017 example: none. Matt Ryan in 2016 scored at this level, though.

* 3000 TSP (7.5 CV) is a legendary season, and the player always wins MVP. There have only been seven, the most recent being Peyton Manning in 2013 and Tom Brady in 2007.

I'll begin with raw data: QB-TSP for the top 40 in passing yards from the 2017 NFL season. The era-adjusted score, in the final column, is the one that aligns to the categories above.


You can work out the math yourself if you're interested, but next week I'll break down why Alex Smith ranks first. In this article, what I really want to present is a different way of using TSP. This is a supplement, not a replacement, to the methodology I've published in the past. It derives from a belief that we get a fuller understanding of issues by looking at them from more than one angle.

Each season, TSP produces a numerical value for quarterbacks. They are easily sorted according to these values, as shown above for 2017. It is therefore easy to compile a list of, for instance, every top-ranked QB by year, beginning in 1946: Otto Graham, Otto Graham, Tommy Thompson, Otto Graham, Norm Van Brocklin, Bob Waterfield, Otto Graham, Otto Graham, Norm Van Brocklin, Otto Graham, Tobin Rote, John Unitas, Bobby Layne, John Unitas, Milt Plum, Sonny Jurgensen, Y.A. Tittle, Y.A. Tittle, John Unitas, John Brodie, Frank Ryan, Sonny Jurgensen, Daryle Lamonica, Daryle Lamonica, John Brodie, Roger Staubach, Joe Namath, Roman Gabriel, Ken Anderson, Ken Anderson, Bert Jones, Roger Staubach, Roger Staubach, Roger Staubach, Brian Sipe, Dan Fouts, Dan Fouts, Joe Theismann, Dan Marino, Joe Montana, Dan Marino, John Elway, Boomer Esiason, Jim Everett, Warren Moon, Mark Rypien, Steve Young, Steve Young, Steve Young, Brett Favre, Brett Favre, Brett Favre, Steve Young, Kurt Warner, Jeff Garcia, Kurt Warner, Rich Gannon, Peyton Manning, Peyton Manning, Peyton Manning, Peyton Manning, Tom Brady, Drew Brees, Drew Brees, Tom Brady, Aaron Rodgers, Tom Brady, Peyton Manning, Aaron Rodgers, Carson Palmer, Matt Ryan, and Alex Smith.

Of course, a lot gets left out of that. Fran Tarkenton, for instance, never led in TSP, but he ranked 2nd or 3rd nine times. There are three specific calculations I keep with regard to yearly TSP ranks:

1.Number of Top-10 seasons. This is a statistical attempt to measure consistency of high-level performance. In the 1940s and '50s, when the league was smaller, I only honor Top-5 seasons.

2. Top-10-Points. This is a simple calculation in which 1st place is worth 10 points, 2nd place is worth 9 points, and so on down to 10th place worth 1 point. This recognizes that not all Top-10 seasons are created equal. It's a much more significant measure than the previous category. In the '40s and '50s, 1st place is worth 10, 2nd is worth 8, 3rd is worth 6, 4th is worth 4, and 5th is worth 2.

3. Year-Points (or yoints, if you're not into the whole dignity thing). Multiply Top-10 seasons by Top-10-Points.

All of these methods are quick-and-dirty. They offer advantages and drawbacks relative to Adjusted TSP and Career Value.

* The statistics above — Top-10 seasons, Top-10-Points, and Year-Points — ruthlessly cut through era differences. There's no subjective weighting (beyond the basic TSP formula), no 5-year average. Players are compared directly and exclusively to their peers in the same season. Players from the early '60s are at an advantage, because the league had only 22 teams, and those since 2002 at a slight disadvantage, in a league with 32 teams, but most of the differences are trivial. This method cuts through trends in strategy, scheme, etc., to evaluate players solely against their most direct peers.

Conversely, however, excellent seasons like Tom Brady's 2007 and Peyton Manning's 2013 rate the same as Carson Palmer's 2015 and Alex Smith's 2017. Bart Starr and Tony Romo are among the players who look better by this method than by my standard approach.

* Low-level compiling does not register in this system. Any season outside the top 10 counts as zero. Last season, for instance, Kirk Cousins scored the same zero as Eli Manning, DeShone Kizer, or John Elway.

Players like Vinny Testaverde and Steve DeBerg rate poorly using this approach, in which longevity and sustained production count for less than an impactful prime. If all you care about are a player's really good seasons, Top-10-Points and Year-Points can be an effective way to measure those. Roger Staubach, whose TSP value is derived almost exclusively from his prime but who had no statistically exceptional seasons, rates very well by this approach.

* While none of these three statistics is a substitute for Career Value, Top-10-Points is probably the most meaningful of the three. Year-Points are interesting, but vulnerable to high-level compiling. I'll illustrate with Jim Hart and Joe Namath. They were roughly contemporary, born less than a year apart. Namath played from 1965-77, Hart from 1966-84, though his last top-10 season was 1978.

Namath had six top-10 seasons: 1st, 2nd, 4th, 5th, 7th, and 8th. That's 39 Top-10-Points, tied for 23rd all-time.

Hart had eight top-10 seasons: 3rd, 4th, 5th, 6th, 7th, 8th, 9th, and 10th. That's 36 Top-10-Points, 27th all-time.

That's close, with Namath slightly ahead because he had a stronger prime. His best season (1st) is better than Hart's (3rd). His second-best season (2nd) is also better than Hart's (4th). His third-best is better (4th vs. 5th), as is his fourth-best (5th to 6th). At fifth- and sixth-best, they're tied. Hart has 9th- and 10th-place ranks as gravy.

Using Year-Points, Namath scores 6 times 39 — 234. Hart's calculation is 8 times 36; he scores 288. The rationale behind Year-Points is to acknowledge the difference between a Top-10 season and a non-Top-10 season. Most of the latter are not 11th- or 12th-place near-misses, but rather mediocre or even bad seasons. The mere 1 point offered by Top-10-Points fails to acknowledge this, but Year-Points does.

I would argue, however, that it over-corrects the problem. Hart rates 23% ahead of Namath, a significant gap. Hart had more good seasons than Namath, beyond any dispute. But equally beyond dispute is that Namath's best seasons were superior to Hart's. Namath was regarded as the best QB in football, while Hart made four Pro Bowls but never shared that kind of reputation.

We all balance the value of excellence and longevity differently, but I feel that Year-Points tends to overvalue the latter. It's an interesting metric with some valid applications, but as a measure of careers, it's a poor substitute for Career Value. Like Top-10-Points, it functions best as a supplement to CV.

Over the next seven weeks, I will revisit my analysis of the top 100 quarterbacks of all-time. If you're interested in that series, I strongly recommend that you read or re-read both my 2015 series on the greatest quarterbacks ever and last year's article on the top-ranked QBs by TSP. Those will provide important context for the coming articles: please review them or check them out. The big QB series begins next Tuesday.

Leave a Comment

Featured Site