There once was a man from Nantucket
When young he held a football and chucked it.
When older, Football had him
So one day he chose, on a whim
To tell Football Convention to suck it.
By now, we know the story. When Chip Kelly presided over the University of New Hampshire football program (he’s not from Nantucket, but close enough), he struck a friendship with Bill O’Brien, who coached at Brown. The two men have shared meals, drinks, laughs, and maybe a secret handshake or two, but more importantly, they talked a lot about football. It was this connection that brought Kelly, the Oregon coach, to Foxboro to see O’Brien, the New England Offensive Coordinator, who then introduced Kelly to Patriots head coach Bill Belichick.
Fast forward to the 2012 NFL season, Week 5, in a game littered with Chip Kelly’s fingerprints. A blistering-quick no huddle offense paced the New England Patriots to a 31-21 victory over the visiting Denver Broncos. The Patriots ran 89 offensive plays, second most in franchise history, and set a franchise mark in first downs with 35. According to Boston Globe writer Greg Bedard, this one game serves as a fine example of the impact Kelly has had on Belichick and the Patriots’ offense. It also serves as a good case study for what the Philadelphia Eagles expect to become. In that regard, much has been written about how Chip Kelly, either through his offense or coaching technique or whatever, will change the NFL and revolutionize the game of football. Revolutionize, he won’t. However, more simply and rather elegantly, Kelly will be executing a strategy. But is it a strategy that will work?
There are some who claim that a byproduct of running a faster paced offense is a defensive unit that will spend more time on the football field. By this logic, a team that wishes to play fast offensively will also need to employ a defensive unit that is deeper and/or better conditioned than the average NFL defense. However, while still a nice luxury, this may not be the case. During the New England Patriots win over the Denver Broncos last season, the Patriots’ 89 offensive plays encompassed 36 minutes of time, leaving the Patriots defense on the field for 24 minutes (against Peyton Manning). Granted, all games are not equal, and it would be unfair to transpose this one game onto what the Eagles hope to accomplish over time. Yet, if we compare the number of offensive plays to defensive plays in games during the 2009-2011 seasons, we see a mild and statistically significant relationship. As the number of offensive plays increases, the number of defensive plays decreases. This could indicate that poorer defenses facilitate more offensive plays. Or it could not (a poorer defense could allow an offense to score using less plays, who knows). At any rate, it appears, at least historically, that the Eagles’ defense may benefit from a faster paced offense that runs more plays than average. (Obviously, this point becomes moot if their defense routinely executes "three-and-outs".)
When using total offensive plays to measure the speed of an offense (that is, how fast the offense operates and executes), we can glean how successful the "play fast" strategy has been. From 2009 to 2011, teams that run more offensive plays than their opponents win 58% of the time. Over a sixteen game season, this translates to a 9-7 record; not exactly indicative of a full-proof strategy (again, it could also illustrate that all defenses are not created equal). But this is a case where a descriptive statistic does not tell the whole the story.
In order to really maximize the effect a high number of offensive plays has on the outcome of a game, let’s assume (based on above) that Chip Kelly just isn’t just interested in executing a shitload of offensive plays; he’s equally interested in executing a shitload more than his opponent. Unfortunately, according to linear regression, this strategy does not necessarily equate to scoring more points than an opponent. From 2009 to 2011, there is only a slight positive relationship (if any at all) between team offensive play differential and point differential per game.
Logistic regression, on the other hand, tells us something else. Using this model, we can determine historically to what degree running more offensive plays than an opponent has helped win games, and project the impact going forward. And, as it turns out, different degrees of offensive play differential has indeed had an impact on game outcomes. From 2009 to 2011, teams had a 50% chance of winning when their offensive plays are equal (actually, this behavior favors the home team… more proof of Home field Advantage?). For each additional offensive play above 50%, a team gives itself a 2.65% better chance of winning. For every ten plays above 50%, a team gives itself a 35.1% better chance of winning; every twenty plays, a 49% chance of winning.
We can retroactively apply these models to the New England/Denver game, where the Patriots ran 23 more plays than the Broncos (89-66). According to the 2009 to 2011 linear and logistic models, this translated to the Pats having a 65% chance of winning the game (70% at home) by 8.2 points (they won by ten, 31-21).
Moving into the future and shifting back to the Philadelphia Eagles, let’s assume that Chip Kelly would like to average 20 more offensive plays than his opponents. This would translate to a 63% chance of winning (68% at home, 57% away), by an average of six points (seven at home, five away). Applying this strategy (if executed) over a 16 game season, the result could be a 10-6 record. However, I think it might be safer to conclude that a team which averages ten more offensive plays than its opponents can attain one more win than if offensive plays were equal; a team that averages twenty more offensive plays can attain two more wins, and so on. Despite their 89 play effort against Denver, the 2012 New England Patriots averaged nine more offensive plays than their opponents, so one of their twelve wins can be attributed to this differential.
Based on these results, it does not appear to me that a revolution is on the horizon. However, Chip Kelly will be implementing a strategy, and a sound one at that. History tells us if Kelly’s offense averages eighty plays per game (twenty more than his opponents) his defense will be on the field for six plays less per game than the league’s season average and his team will have a 49% better chance of winning each game. In this regard, playing fast does seem to offer an advantage. Yes, intentionally increasing offensive plays has not been the typical NFL convention. But in the end, Chipper doesn’t really care about that, does he.