Tuesday, May 15, 2018
New Stat: Game Manager Index
All American football fans recognize the term game manager, referring to a modestly talented passer who takes few chances, hoping to ride a ground game and defense into the postseason. What he lacks in electricity and talent, he makes up for with caution and intelligence. Such players can be a good fit for excellent teams who win games on the strength of their defense and a sound running game. Those teams need a quarterback who avoids mistakes more than they need one who leads a brilliant two-minute drill.
On most teams, though — those without a historic defense and a great running back — game managers are a mild curse, frustrating players who make sure you go 8-8 every year. The term game manager is a pejorative, implying that luck, more than skill, determines the team's success. A game manager simply must play for an excellent team.
I recently developed a formula, Game Manager Index, to identify game managers. The calculation includes completion percentage, yards per completion, touchdown percentage, interception percentage, and net yards per attempt. Game managers play conservatively, so they tend to record high completion percentages and low INT rates. They sacrifice average yardage, though, and they tend to rely on the ground attack and kicking game for points. Often, their teams don't score that much anyway, winning games 17-13 rather than 35-31.
Anyway, here's the formula: [ 175 + (.75 * comp%) - (10 * yds/comp) - (4 * TD%) - (10 * INT%)) - (4 * NY/A) ] * 1.5
Everyone starts with a score of 175. To that, we add 3/4 of completion percentage, then subtract 10 times the yards per completion and interception percentage, and four times the touchdown percentage and net yards per attempt. When evaluating careers, this typically produces a Game Manager Index (GMI) between 25-75. Higher scores imply a game manager, while low scores imply a gunslinger.
Score vary more for individual seasons, but normally remain between 0-100. This is not a hard limit, though. In 2017, two qualified passers scored over 100 (Joe Flacco and Tyrod Taylor), while Deshaun Watson (although he didn't play quite enough to qualify) would have rated at -27.6. There are many exceptions, but in general, great QBs score below average on GMI without reaching extremes like Watson. Most often, a great QB tends to rate in or near the 40s.
Let's walk through some QBs you probably remember and show what the GMI formula tells us. For each decade below, I calculated GMI for every passer with at least 10,000 passing yards in the decade.
From 2010-17, the highest GMIs belong to Sam Bradford (89.4), Alex Smith (80.7), and Derek Carr (78.4). The lowest are Cam Newton (33.7), Jameis Winston (35.2), and Andrew Luck (39.8). Tom Brady and Drew Brees both score 59.5, with Aaron Rodgers at 50.3. At a glance, those figures seem intuitive and tenable. No one would describe Newton or Luck as a game manager, while everyone would agree that the term fits Bradford and Smith.
From 2005-14, the highest GMIs are Jason Campbell (73.3), Alex Smith (72.7), and Marc Bulger (68.4). The scores are lower for this time period because short passing has taken over the NFL. By the standards of the past, everyone is a game manager now. The lowest GMIs for this decade belong to Eli Manning (32.8), Cam Newton (34.2), and Jake Delhomme (37.4). Tony Romo (37.6) and Philip Rivers (39.7) also rate very low on GMI. Brady and Brees score about 53, with Peyton Manning at 50 and Rodgers at 42.
In the 2000s, the highs are Jason Campbell (78.9), Brad Johnson (77.6), and ... David Carr (75.1), which I admit seems a little weird. David Garrard, Rich Gannon, and Chad Pennington are the next three, though, which hopefully restores some confidence. Maybe Carr played differently than I remember, or maybe his stats are just skewed by how bad the expansion Texans were in the early 2000s.
The lowest GMIs of the decade are Tony Romo (19.0), Ben Roethlisberger (21.9), and Trent Green (27.8). Peyton Manning comes in at 42.5, Tom Brady at 53.5, with Brett Favre at 48.8 and Drew Brees and Donovan McNabb both about 56.
From 1995-2004, the statistical game managers are Gannon (67.5), McNabb (65.8), and Troy Aikman (64.5). The gunslingers are Kurt Warner (15.7), Chris Chandler (25.8), and Gus Frerotte (27.4). Favre and Manning, the dominant QBs in this time period, rate 32.9 and 32.4, respectively.
In the 1990s, the highest GMIs are Neil O'Donnell (67.6), Troy Aikman (61.9), and Jim Harbaugh (60.5). O'Donnell is perhaps the quintessential game manager, so it's reassuring that he rates atop the list. The lowest GMIs of the '90s are Vinny Testaverde (19.4), Chris Miller (21.4), and Jim Kelly (21.5). Aikman and Kelly represent opposite ends of elite spectrum, with Aikman performing in the Bart Starr/Bob Griese model as a high-level game manager on a great team, and Kelly throwing caution to the wind by comparison, his offense living or dying with the pass.
Other elite passers of the '90s: Warren Moon (45.2), Dan Marino (43.8), John Elway (40.3), Steve Young (38.2), Brett Favre (38.0).
The GMI formula is designed for the West Coast Offense era, so I didn't calculate it for any decades prior to the 1990s. This is just trivia, obviously. It doesn't tell you whether a player is good or effective; if you want a stat to identify the best and worst quarterbacks, please refer to QB-TSP. What GMI can do is illustrate which passers tend to play conservatively, which ones are game managers and gunslingers.
In the '90s, for instance, Aikman and Kelly were both great QBs, but they played with very different styles and succeeded in pretty different ways. Aikman, despite his ability, was seldom called upon to win the game, since Emmitt Smith and the Dallas defense often had things well in hand. Aikman's duty in those games was to play like a game manager, to avoid turnovers and not to lose a game his team was controlling. Kelly, in contrast, usually played with average-to-poor defensive units, and had to put points on the board to pressure opponents. Give Buffalo a double-digit lead, and all of a sudden Bruce Smith and Cornelius Bennett could tee off on opposing passers. Kelly played more aggressively because that was what the situation called for.
Could Kelly have been a great quarterback in Dallas, and Aikman with the Bills? Almost certainly. In other cases, a player's tendencies are dictated by his mentality or skill set. Sam Bradford has been a game manager everywhere he played. Neil O'Donnell and Brad Johnson always played pretty conservatively; they didn't visibly adapt their approach to the teams they played on.
Neither style — game manager and gunslinger — is inherently more effective than the other. However, given an effective QB with a high or low GMI, I'd have more confidence that the low-GMI passer is truly a great player, rather than a product of his circumstances. I also think that low-GMI players are obviously more fun to watch than high-GMI players.
Do with this stat what you will, but don't read too much into it. GMI is a fun way to categorize playing styles, but not a reliable method to assess ability or quality of performance.