Hey BGN, how's everyone holding up in the offseason? I've been quiet around here recently mostly because the past two months have been a blur of family visiting, moving to a new place, and vacation. My vacation took me through Philly for a few days, and I made the most of it by grabbing a cheesesteak at Jim's on South Street and getting the obligatory photo at the Rocky statue, among other touristy things. I hope at least keeping clear of Pat's and Geno's is enough to make you locals happy.
But I digress.
For those of you who may have joined the site recently, each year I try to come up with a purely statistical formula that can measure the strength of a professional football team, which I call "Crunching The Numbers." Wins are not factored into the equation and I watch exactly zero minutes of tape. The idea is to find the "magic metrics" that can separate a 'pretender' from a 'contender' as early on in the season as possible. It remains to be seen if such metrics actually exist, but if they do I'll find them. Eventually.
Each year I try to stay current with NFL statistical trends by analyzing different metrics. That way I can steadily improve on my formula over time. In years past this was accomplished by determining the correlation coefficient between regular-season wins and various metrics, but after some good constructive criticism last year I am trying a new approach. Instead of throwing wet spaghetti at the wall and seeing what sticks, I will first develop and test several hypotheses on what I believe would make a successful NFL team. Then, based upon the evidence provided I will develop new hypotheses and test those. This process will iterate until I have developed what I believe to be a good model for the upcoming season.
Today's post will introduce my first round of hypotheses and the results of the tests will be discussed tomorrow. Don't worry, I'm not going to do a post on each iteration. You'll see the final results when I debut this season's version of the formula later in the offseason.
Below I have listed twelve total hypotheses with a brief rationale. There are six each for offense and defense:
A successful offense has...
High yards per pass attempt. There really shouldn't be too much surprise with this one, as the impact of YPA on success is well documented. In the past I've used the full passer rating but I've since realized there are too many collinear variables in that metric for it to be a viable option. However, to exclude YPA from analysis purely because we are confident in the outcome would be bad practice.
A low interception percentage. Again, nothing very groundbreaking here. The ability of a quarterback to avoid mistakes (especially dumb ones) is instrumental in any team's long-term success. Just ask Mark Sanchez.
A low sack rate. Pretty straightforward with this one. Teams that can keep their quarterback upright, either with a great line or a mobile passer, are more likely to succeed down the stretch.
Low fumbles per game. This also falls into the "avoid mistakes" category, but I think there is a slight emphasis on discipline here. Good players can avoid fumbles by protecting the ball and not being careless does take a level of discipline. I'm a big believer that good teams are well-disciplined and that's why included this.
A high number of rushing first downs per game. The idea here is two-fold. One, I am going to offer the idea that a rushing first down is indicative of a team getting itself into ideal third-and-short situations on a regular basis (I tried to find rushing third down conversions but it wasn't being tracked). Two, once a team has a lead it will try to kill the clock by running the ball. Good teams will convert first downs even though the defense knows what's coming.
A high scoring rate. Easily the most controversial hypothesis here. I said I wasn't accounting for wins in this formula; I never said anything about points.
A successful defense has...
Low yards per pass attempt allowed. Like the offensive counterpart, this is very well-documented. Nothing scandalous about this.
A high interception rate. You know what the best way to stop an opponent's offense is? Take the ball and give it to yours!
A high sack rate. One of the most frustrating things to watch is an opposing quarterback who has all day to survey the field from the pocket. The world champion Broncos marched through the playoffs by wreaking havoc in the backfield, so I'd say that this metric is as good as any for success.
A high fumble recovery rate. This is probably my biggest reach from the outset. The gut feeling here is that fumble recoveries are mostly luck and shouldn't have an effect on a team's success. But "gut feelings" don't exactly fly as research, so I'm going to test it here. The idea is that a disciplined defense will have the awareness to recover as many fumbles as possible.
A low third down conversion percentage allowed. Ultimately, the defense's job (other than forcing turnovers) is to get off the damn field. Enough said, really.
A low scoring rate allowed. *GASP* More controversy!
One side note: even though it went against my schooling as an engineer, I didn't quantify any of these hypotheses on purpose. This is because football stats are relative - the only "good" number is the one that's better than everyone else's. I would eventually like to establish a baseline of numbers so I CAN quantify these things, but the game evolves so fast that I'm not sure if that is a reasonable goal. It will be something that I keep an eye on.
Check back tomorrow when I reveal the results of the analysis. In the meantime, feel free to discuss your opinions on my hypotheses if you're really that bored at work.