The QB - Laying the Foundation

There would be little argument from most people that the quarterback is the most important player on the football field. They’re typically the team’s leader, or their scapegoat. They get the credit when they lead their team to a thrilling come-from-behind victory. And they take the blame when they cannot drive the field late with the game on the line.

In the last few years I’ve felt we have been seeing a transformation of the position. It seems that more and more very good to great quarterbacks have been popping onto the scene. I believe I wrote somewhere – I think in an article I wrote for Profootball24x7 – that the game used to often come down to if one team had a terrific QB, if the game was close, that QB would drive them the field and win the game while the other could not. But that now, with several teams having great QBs, it would come down to whoever had the ball last, to drive the field and win the game. “Tom Brady trots onto the field, down 4 with three to play. Can he drive the field? And can he take up enough clock not to allow Phillip Rivers the time to drive the field if they score?”

I’ve also been intrigued by a question that I’ve seen popping up more and more in discussions with friends and fellow fans. As Donovan McNabb’s career comes to a close, is he a Hall of Fame quarterback? His numbers are certainly impressive. He’ll finish in the top ten in several very important passing categories. But given the rate at which I have perceived the QB play to have improved over the last decade or so, I wasn’t convinced that was enough.

And so I set out to try to find out exactly how much QB play has improved the last few years, and try to answer the question of whether McNabb is truly good enough for Hall worthiness or not. The result is this series of three posts. The first - this one - will talk about the methodology and reasoning behind my analysis of the QB position. The second will be the actual analysis of the position. The third will be an in depth look at McNabb and his Hall worthiness.

I’ll start with a note that Patrick, Jim and I have had a very interesting email conversation about quarterback play and how to best analyze it. I won’t post the whole conversation, but the general discussion centered on how much we can or cannot use statistics to tell the story of the position. There’s still disagreement on exactly how much of the story can be told using stats, but it seems clear that the answer is some form of “Stats can tell part, but certainly not all of the story.” This actually gets truer the more you look at smaller data sampling. For instance, if you want to look at how much QB play has improved over the past 25 years, statistics will be able to tell a larger portion of that story than they will be able to tell you how well (or not well) Kyle Orton played in 2009.

To truly get the entire picture of quarterback play improvement, we should look beyond the statistics. How has the play of wide receivers improved over the same period? How have NFL offenses evolved in this time? Some of the non-statistical measures will be examined in the next two posts, in addition to the statistical evaluation.

So to start on the statistical analysis, a framework had to be developed. It wound up somewhat arbitrary, as any analysis tends to be. I probably could have developed a much deeper, more precise frame-work for this. But let’s be realistic. It took several hours to pull and cut the data I have for this, and this isn’t my profession. I’m not taking an “80/20 is good enough” attitude. But there’s no realistic way to add additional value to what I’ve done without spending significantly more time on it. Given this isn’t my profession, I can’t really afford that, so this was my best effort.

I began by defining my arbitrary cuts as looking at the following boundaries:
- No more than 25 QBs per season would be looked at.
- The top 25 QBs by QB rating would be looked at.
- QBs must have 200+ attempts on the year to be looked at.
- No seasons prior to 1960 would be looked at.

What was my reasoning behind this? I used QB rating as the barometer because, despite the criticism that seems so fashionable these days, it does encompass the major aspects of how QBs tend to be viewed in their effectiveness. Completion percent, yards per attempt and TD:INT ratio are all considered in the rating, and I feel no other single metric does a better job at evaluating the position. While it’s not a good metric to tell the entire story of the position, when you have to say “Pull QBs by rank of [something],” you either have to pick one metric to rank them by, or you have to do a ridiculous amount of work to combine several metrics. Ironically, QB rating already does this for us. Plus, any new fancy-schmancy ranking I could come up with that blends several different factors could simply be called arbitrary anyway, so I took the path of least resistance since it also seemed to be the path of most sense.

Going only back to 1960 was definitely an arbitrary and somewhat ridiculous cut, but I felt like this was a good enough cut around what defines modern era QBs. The NFL Hall of Fame actually segments between Modern Era and Pre-Modern Era players. Now, there are guys like Y.A. Tittle and Bob Waterfield who played most or all of their careers prior to 1960. But realistically the value in going prior to ’60 wasn’t really high and just seemed to be creating more work.

The cut at 25 QBs was mostly to cut out the fat. Some work could have been done here to find a better cut, perhaps. But while the NFL had 26-32 teams for a majority of the years for which I captured data, not all have starting QBs worthy of being looked at. I didn’t want to cut out mediocre to poor starters. But I did want to cut out the absolute trash, and the guys that either only played because injuries forced them into action or guys who’s play was hurt because of injuries. For instance, there doesn’t seem to be a ton of value in capturing Joey Harrington’s ’06 season and comparing it to Vince Farragamo’s 1985 season.

Finally I used a minimum of 200 passing attempts because that’s what Pro-Football-Reference uses for its statistical qualifier. I went to PFR, went year by year, sorting QBs by QB Rating, capturing the top 25, and dumping it all into an Excel file.

Interestingly, there was a very distinct cut for where I could get 25 players to pull in a season. After 1976, there were always at least 25 players (though sometimes exactly 25) that threw 200 attempts in any given season. Prior to 1976, there was never a season in which 25 players qualified. There was also a pretty distinct cut at 1970. Between ’70 and ’76, an average of 22 qualified with no less than 17 throwing 200 attempts. From ’60 through ’69 – prior to the AFL/NFL merger – an average of 13 qualified with no more than 14 throwing 200 attempts. Because of this, some of my data cuts were taken with this lens.

From this data it was a matter of ranking everything and then pivot-tabling and cutting the data. I added columns and manually sorted by [year] then [data set] to rank every QB by season in every major passing category. Then I started graphing various things I found interesting.

And that’s pretty much it. The methodology behind the madness that’s about to ensue. Is it perfect? Hardly. Not even close. But I think it will provide a good framework for looking at a few major things…
- How much have the best QBs in the NFL improved vs. the best of prior years?
- How much has the ‘average’ QB improved over the years?
- How much have the bottom-dwellers … the guys considered marginal to poor starters … improved over the years?

These were the questions I really wanted to address, and I think you’ll find they’ve been addressed at least somewhat competently in the next day when I post that analysis. Meanwhile, feel free to post any “You’re an idiot for not considering this” commentary below.