Saturday, November 11, 2006

Quarterback Production

It's the midpoint of the '06-'07 campaign, and sportswriters everywhere have submitted their picks for first half MVP, first half defensive player, best rookie, et al. There appear to be two constants among these picks- Peyton Manning as the first half MVP, and Marques Colston as the offensive ROH.

Manning's biggest competition looks like San Diego's L. Tomlinson, who has enjoyed a monster first half of the season. We delve deeper however and ask ourselves a question most have simply assumed the answer to: Has Peyton Manning really even been the most productive quarterback of the first half?

The short answer? No.

Now before getting into why that is, it is important to look at the most commonly used metrics to measure quarterback performance today. While total passing yards, passing completion percentages, touchdowns, and interceptions are certainly looked at individually, the one statistic that is most referenced in discussions about quarterbacks is passer rating, or RAT+. (+ Ratings are basically the original metric adjusted for a specific year so that 100 would be the average.)

At a glance, QB rating seems like a perfect metric; it takes into account number of passes, completions, yards, touchdowns, and interceptions. Most importantly, year after year, "good" quarterbacks have "good" ratings, and "bad" ones don't. It has two crucial flaws however:

1. Each team functions based on a specific type of offense it generally employs for the course of an entire season, or more generally for the course of its head coach's tenure (West Coast, etc). Some systems are more conducive to the running game than the passing game; in these systems quarterbacks don't gamble as much by throwing deep passes and, therefore, limit themselves to the short passing game. The short passing game is obviously far easier for a quarterback to play than the deep one, as far fewer risks are involved.

A quarterback would have a far easier time succeeding in this system and posting a specific passer rating, than a quarterback who plays in a deep pass oriented system, simply because his offense relies more on his running back than on his passing arm. Passer rating does not even remotely take this into account; indeed, it looks at attempted/completed passes, but it does not penalize a quarterback who has fewer attempts than another quarterback since it simply takes the ratio between them.

2. Passer Rating does not take into account the amount of pressure a quarterback receives per play. The most clear example of that in the NFL today would be Peyton Manning, who receives tremendous pass protection from his offensive line, and a quarterback with similar talent Carson Palmer, who does not. Is the difference in their passer ratings entirely due to this? Of course not, but a quarterback who posts similar statistics to another quarterback who does not see as much pressure on average should be considered more productive.

Continuing then, there exist five major components to a quarterback's production. First is his ability to get the ball into the end zone, measured obviously by Touch Downs. Second, is the amount his offense leans on the passing game; this is measured his attempts per game relative to the total number of plays run, as well as his yards per completion average. Third, is the quarterback's ability to throw a large volume of passes his opponents do not catch. If there is one statistic passer rating undervalues, it is interceptions. Fourth is his ability to perform under pass pressure, measured crudely by the number of times he has been sacked. (If the NFL published "hurried" statistics, this would be a far better metric). Finally, his ability to stay on the field uninjured impacts his overall production.

Keeping these five things in mind, I came up with a better version of QB passer rating that includes one section for each of the four first items multiplied by the number of games the player has played divided by eight (number of games into the season.) I first checked to make sure the equation didn't provide random values for players, and that no one section of the equation could overshadow the others. (I weighted the "pressure" part the least). The results were pretty interesting, and upon closer inquiry, it is clear that they are not wrong.

First, here they are:

Marc Bulger


Peyton Manning


Donovan McNabb


Damon Huard


Philip Rivers


Drew Brees


Mark Brunell


Brett Favre


Carson Palmer


Jake Delhomme


Tom Brady


David Carr


Jon Kitna


Eli Manning


Bruce Gradkowski


Rex Grossman


Alex Smith


J.P. Losman


Michael Vick


Jake Plummer


Brad Johnson


Steve McNair


Chad Pennington


Charlie Frye


Byron Leftwich


Matt Hasselbeck


Tony Romo


Andrew Walter


Matt Leinart


Drew Bledsoe


Ben Roethlisberger


The first reaction is clearly "Marc Bulger... Peyton Manning... what!?!" It is necessary, though, to examine what exactly in the statistics made for such an order in the ratings. First section: scoring. Manning owns this 17-13. Nothing too interesting here. Second section: How much a team relies on its passing game. The average football fan would say that the Colts rely more on their passing game than the Rams. The statistics say no, as Bulger averages more attempts per game than his Indy counterpart. However, this is a pretty minor difference, and Manning owns the second part of the second section by averaging more yards per completion. What gives? Aha. The third section: interceptions. Bulger has, ridiculously, tossed just one interception in 290 attempts. Manning? Three over 281. It may not look it, but that's a huge difference; it's the difference between a ratio of 290 and 94. Bulger owns a big advantage in the fourth section as well, as he's been taken down 24 times as opposed to Manning's 10. To sum, the stats are nearly identical (2300-2291, 64.1-65.1, 290-281, 13-17, 1-3) but Bulger has been slightly less turnover prone in an offense that throws the ball a hair more than the Colts, and behind an offensive line that's making life for Bulger pretty hairy.

Observation Number Two: Brett Favre has been having a far better season than most people are giving him credit for. True, he's tossed seven interceptions, but keep in mind that Green Bay is relying heavily on the pass; Favre's tossed far and away the most passes in the NFL with 310.

Observation Number Three: Damon Huard's not nearly as close to Peyton Manning in terms of talent/production level as Passer Rating would have you believe (Huard ranks a 105.2 to Manning's 106). A lot of Huard's success can be credited to the fact that he's got L. Johnson to hand the ball off to as he's barely broken the 200 attempt mark on the season. Still, he's pretty good. Just not that good.

Observation Number Four: Drew Brees has been good, but not as good as his replacement in San Diego, Phillip Rivers. Rivers, indeed, also has L. Tomlinson to hand off to, but he's proven that he's more than capable of playing more than a short yardage game. His 7.34 Y/C is just under Brees' 7.67. Brees' major shortcoming is his INT/ATT ratio.

Observation Number Five: Ben Roethlisberger has been the worst QB in the NFL. That's right. Worst. This is where the ability of this formula really starts to show. Roethlisberger has had basically the same season as Jake Plummer, according to passer rating (75.5 to 75.2). I'm pretty sure Pittsburgh and Denver fans alike would beg to differ.

So... is Peyton Manning really the MVP so far? Well, he does have eight WINS...

No comments: