Tuesday, May 30, 2017
The Worst Quarterback of 2016
One of the pro football sources I read regularly is Rotoworld. It includes a lot of fantasy football advice, but it's also a reliable place to find news about player contracts, injuries, and general league news. This article is not about Rotoworld, or something silly that I read there recently; it's just a jumping-off point.
A recent post at Rotoworld stated, "[Ryan] Fitzpatrick was the worst quarterback in football last season." That's absurd, and obviously untrue. But what interested me wasn't the sloppy misstatement, it was the question: who was the worst quarterback in football last season?
When we talk about the worst QB in football, I think we're only talking about guys who really played. I mean, Packers backup Brett Hundley had a 0.0 passer rating ... on 10 attempts. He went 2-for-10 with 17 yards and an INT. But Hundley didn't play enough to evaluate him fairly, and he didn't really do any harm to his team. When we talk about the worst QB in football, we're looking for regular starters, guys who went out there week after week and stunk it up. Fans weren't tearing their hair out over Hundley. They were tearing their hair out over Brock Osweiler and Blaine Gabbert and Fitzpatrick.
Let's begin with Fitzpatrick, even though — as mentioned above — he certainly was not the worst QB last season. He wasn't good. Fitz went 223-of-408 for 2,710 yards, with 12 TDs and 17 INTs. That's a 69.6 passer rating, though his low sack rate salvaged a 6.2 NY/A (net yards per attempt). He got to play with Brandon Marshall, who had the worst season of his career, but I'm not sure that's entirely Fitzpatrick's fault; Bryce Petty was even worse: 5.0 NY/A, 60.0 rating. Maybe Petty is ineligible for "worst QB" mention, since he only started about a quarter of the season, but it seems indefensible to me to call Fitzpatrick the worst quarterback in football when he clearly wasn't the worst quarterback on his own team.
Fitzpatrick's -5 TD/INT differential was second-worst in the NFL, just ahead of Chicago's Matt Barkley (-6), and his 2.0% fumble rate was high. But Fitzpatrick generated yardage and first downs much more efficiently — 6.2 NY/A and 30.3 first down percentage — than the quarterbacks who were really the NFL's worst.
Consider, for example, the Texans' Brock Osweiler. He went 301-for-510 with 2,957 yards, 15 touchdowns, and 16 interceptions. His 72.2 passer rating was better than Fitzpatrick's, he had fewer fumbles (9-5), and he scored two rushing touchdowns. But Osweiler averaged 5.1 net yards per attempt. That's more than a yard behind Fitz. One yard per play, over four or five hundred attempts — actually Osweiler had 537 — is an enormous difference, worth much more than a handful of touchdowns and turnovers.
The league average for net yards per attempt was about 6.4. Osweiler, in 15 games, averaged over 6.4 NY/A once. He topped the league average once in 15 games. In Week 7, he set the all-time record for fewest yards with 40 or more pass attempts. The Jets were terrible last year, with or without Fitzpatrick, but Osweiler visibly held Houston back from — potentially — contending for a championship. Fitzpatrick made too many mistakes, but Osweiler committed a more fundamental sin: he completely failed to generate any offense.
Three Bears passed for over 1,000 yards in 2016. One of those quarterbacks played well (Brian Hoyer, 98.0 rating) and two are in the "worst QB of 2017" conversation. Matt Barkley, as mentioned, had the worst TD/INT +/- in the league (8 TD, 14 INT). His average yardage was good, though (7.1 NY/A). Jay Cutler had a rougher time, with 4 TD, 5 INT, and 6.2 NY/A. Put that yardage figure in context: playing on the same team, with the same supporting cast, Barkley averaged 7.1 NY/A and Hoyer averaged 7.0; Cutler, at 6.2, came in nearly a yard behind. He also had an appalling fumble rate (3.8%), much worse than Barkley (1.7%) or Hoyer (1.4%).
Chicago averaged 20.3 points per game when Barkley started, compared to 16.6 with Hoyer and 14.8 with Cutler.
The Browns' story is similar. Cody Kessler was okay (92.3 rating), but Josh McCown and Robert Griffin III struggled badly. Kessler averaged 5.7 NY/A, McCown 5.3, and Griffin 4.4. Only Kessler had more TDs than INTs. McCown had the worst fumble rate (3.7%), whiile Griffin did contribute as a rusher (190 yards, 2 TDs, 6.1 avg).
Griffin's 4.4 NY/A was the second-worst of any player with at least 100 attempts. The only mark worse belonged to Los Angeles Rams rookie Jared Goff, who averaged a preposterous 3.8 net yards per attempt. That's the lowest NY/A since 1992. The worst mark of 2015 was nearly a yard better (Zach Mettenberger, 4.7). As recently as 2013, no one averaged less than 5 NY/A. Blaine Gabbert as a rookie easily beat Goff's mark, averaging 4.2 NY/A. JaMarcus Russell in 2009 averaged 3.9 NY/A; Ryan Leaf as a rookie averaged 4.3. When Kyle Orton was a rookie in 2005, and couldn't seem to generate any offense, he averaged 4.2 NY/A. Goff's inefficiency was historical. When Stan Gelbaugh averaged 3.6 NY/A a quarter-century ago, he did so without the benefit of the illegal contact rule, the defenseless receiver rule, and the rules that protect the quarterback's head and legs.
We might cut Goff a break if he had been efficient in other areas — but no luck there, either. He threw 5 TDs and 7 INTs, with a 3.4% interception rate more than 50% above the league average. Goff rushed for 16 yards, which is nothing, and scored a rush TD, but he also fumbled five times. Some of Goff's problems are related to the Rams' roster and coaching staff, but let's stick with statistics for now.
I ran fantasy scores for some of last year's least effective QBs. Fantasy scoring isn't uniform, but I calculated (Pass Yds/25) + (Rush Yds/10) + (Pass TD * 4) + (Rush TD * 6) - (INT * 2) - FUMB. That's not a good scoring system, but it's close to what most leagues use. Fitzpatrick scores 126.4, a per-game average of 11.5. His teammate Petty scored 31.3, or 7.8 per game. Osweiler scored 166.4 (11.9/gm). Cutler's at 9.0/gm, and Goff trails the pack with 7.5/gm.
Fantasy scoring underrates yardage and overrates TDs, which is why Fitzpatrick trails Osweiler, but I think the "worst quarterback in football last season" comes down to Brock Osweiler and Jared Goff. Osweiler was more efficient (5.1 NY/A, 72.2 rating) than Goff (3.8 NY/A, 63.6 rating), but he played a full season, taking 567 snaps. Goff only started half the season after taking over for Case Keenum; he handled 239 snaps.
But Goff was so inefficient, I think he surpasses Osweiler. I use the stat QB-TSP, Quarterback Total Statistical Production, to evaluate QBs. Last year, Matt Ryan led the league, 2616. Alex Smith (1050) and Sam Bradford (984) were about average. These are the bottom 10, among players who attempted at least 100 passes:
Case Keenum, LA, 195
Ryan Fitzpatrick, NYJ, 176
Blaine Gabbert, SF, 95
Matt Barkley, CHI, 91
Jay Cutler, CHI, 34
Brock Osweiler, HOU, 0
Robert Griffin III, CLE, -37
Josh McCown, CLE, -98
Bryce Petty, NYJ, -156
Jared Goff, LA, -422
Goff is alone.
TSP is based on linear weights. Quarterbacks score points for things like yards and touchdowns, losing points for sacks and turnovers, but they also lose points on each attempt, so that no one can score well just by throwing a lot: the system assumes that your passes should accomplish something more often than not. Last season, the attempts penalty was 4.5. Thus, an incomplete pass would score -4.5 A four-yard gain would score -0.5. A five-yard gain would score +0.5. A 10-yard gain would score +5.5, and so on.
If I raised the penalty to 5.0, Osweiler would score -291, but Goff would drop to -500. With the penalty set at 5.5, Osweiler scores -631, even worse than Goff's -563. That's an unreasonable standard, I think — 5.5 adjusted yards per attempt. By that measure, Joe Flacco and Blake Bortles would rate as negative-value players. Neither had a particularly good season, but both were viable starters.
So I don't think there's any question: Goff was the worst quarterback in football last season, statistically. But Goff didn't play with DeAndre Hopkins and Lamar Miller; he played with a clueless coaching staff, a hopeless offensive line, and a receiving corps that hadn't produced a 1,000-yard receiver since Torry Holt in 2007.
Brock Osweiler was bad last season. TSP scores him at precisely replacement level, which implies that unsigned free agents could be expected to match his performance. That's bad. But Goff was a disaster. Osweiler was held to a high standard, because he had name recognition and an expensive contract, and he played for a division-winning team that made several national TV appearances. Goff played on a team of which not much was expected, and that was out of the playoff race by the time he stepped in as starting QB. I think if Goff had been on television week after week, like Osweiler was, checking down on every play, people would appreciate what a monstrously bad season he had. Goff, not Osweiler — and certainly not Ryan Fitzpatrick — was the worst quarterback in football last season.
There have been plenty of QBs who rebounded from bad rookie seasons to have fine careers. Terry Bradshaw threw 6 TD and 24 INTs before going on to a Hall of Fame career. Drew Brees and Donovan McNabb underwhelmed in partial seasons to begin their careers. So without trying to condemn his future, I'm content to pass judgment on the past: Jared Goff was the worst quarterback in the NFL last season.