Hey everyone! Over the next year, I want to understand the current state of football analysis and hopefully use that knowledge to start innovating in the space. So while this is crazy late in the season, I want to start writing a weekly post about some of the things I have been looking at. The first thing I have looked into (and the main insight of this post) is Football Outsider’s DVOA. What I wanted to do was create a proxy of these numbers using data that I collected, cleaned, and summarized myself. I still believe that the DVOA ratings have more bells and whistles and thus are more accurate, but hopefully over the next few weeks, I can add some novel insight to these numbers.
Another thing that I want to do is create a “hangover” post where I basically document what I got wrong and what I got right in my predictions. One of the things that I hate about a lot of sports writing is the lack of accountability to numbers and/or predictions that are produced. I’m pretty sure the fundamental idea of learning is making mistakes, understanding them, and learning from them. So having said that, if you watch the games and think “man, that dude is an idiot and got this really wrong” then feel free to comment on this post or let me know in a tweet @pspitler3. I welcome the criticism.
So the numbers that I present are standardized efficiency numbers over expectation. To determine efficiency I defined a play “success” as a first down earning 40% of the yards to gain, a second down earning 50% of the yards to gain and a third or fourth down gaining the yards to gain or more. Thus a team’s total ability to meet these metrics over their total plays defined that team’s efficiency. To account for defenses or offenses that they faced I took the rest of the teams on the competing team’s defenses or offenses schedule and determined what the expected efficiency would be based on that team’s defense/offense. So the difference from expected determined how far above the team was from expected results in playing who they did. The results can be seen here:
These results are pretty interesting in that they fit the narrative of the season quite well in that teams either have elite offenses or defenses, but not both. The teams that do have both in the cloud of SEA/GNB/CIN/TEN/NWE seem to be teams that are favored to win their division with the exception of CIN (and maybe TEN). These teams seem to accurately represent the most “well rounded” teams in football. With a few exceptions (RAI/PIT/DET) It looks like the offensive efficiency numbers work well. DAL/NOR/ATL/WAS are some of the best offenses in the league this year. A future step for this might be adding in expected yards per success and how that correlates to efficiency. The idea behind this is the Raiders seem to pick up huge chunks of yards when they are successful. With the exception of JAX, the defensive efficiency numbers also seem to make sense with the best defenses. One reason for JAX might be the game situation; If a team is up by two scores a “success” is defined more by shortening the game (burning clock) than by gaining yards.
To break this down a little further I wanted to see what the most efficient offenses and defenses were from a pass/run perspective.
This is interesting because it shows Dallas as by far the most efficient offense. They excel in both the run and pass game. NO/WAS/ATL round out the most efficient offensive units. On the defensive side of the ball it seems like teams are built to stop the run or the pass, but not both. Denver has an extremely efficient pass defense and Baltimore has an extremely efficient run defense. The Giants have the overall most efficient defense. They are also the only team to beat the Cowboys and make their offense look anything short of unstoppable. That leads me into…
Things to Watch:
- Cowboy’s offense versus Giant’s defense – In what is the most interesting matchup of the weekend we get to see the most efficient defense play the most efficient offense. If I had to pick, I choose the Cowboys to win mainly because the team that they were in the first game is probably not reflective of the unit they are today. Their quarterback and running back are both rookies that have matured into their roles and I also believe that offensive line play improves as the season progresses due to the limited contact allowed in the offseason. However, if you’re expecting a blowout, I don’t think that is what we will see.
- Baltimore at New England – How one dimensional can the Ravens rush defense make the Patriots? I am curious to see the Ravens rush defense and how that effects the pass/run distribution of the Patriots. The Patriots are notorious for picking on a team’s weakness so I would expect Brady to have a lot of pass attempts in this game. On another note, was Baltimore’s offense last week a blip or the new norm? This one is going to be closer than a lot of people think with a relatively low score. Belichick is a Sith Lord however… So expect the schemiest of the schemes in their path to dominance over the universe.
- Washington’s offense versus Philadelphia’s defense – The closest proxy to Washington’s offense is Atlanta’s offense from an efficiency standpoint. Philly’s defense made the Falcon’s offense look mortal in their matchup. Can they do the same for Washington’s? This matchup comes down to Philly’s rush game against Washington’s rush D. I’m going out on a limb and taking the Eagles at home.
- Atlanta’s defense versus LA’s offense – In the matchup that proves to be a case of the extremely movable object matching up against the supremely stoppable force, what happens? As much as it pains me to admit, Atlanta has one of the worst defensive units from an efficiency standpoint in the league. The upside is they get to face the most terrible offense in the league. Expect the fireworks to be returned to the store in this whamp whamp of a matchup. I expect the Falcons to win this game, but watching Goff look like a real quarterback is going to be annoying.
- Why are the Raiders and Chiefs good? – From a pure efficiency standpoint both of these teams seem to be average or below average. I am extremely interested in looking at this game to understand what I am missing and can add to the efficiency alone to better represent these teams.