Conversely, 1,816 plays were run in the final frame when the offense was up by two or more possessions. Teams called pass 30 percent of the time on those plays.
The above is an extreme example of the importance of considering game situation when evaluating a team’s play-calling.
In an effort to do exactly that, I took a lengthy look at play-by-play data from the past six seasons. I examined NFL-wide and team-by-team pass/run rates during each quarter and split based on the club’s advantage/deficit on the scoreboard. This process will allow us to better project each team’s pass/run rates for an entire season and, once the season rolls around, on a weekly basis. It’s also a very important step in accounting for the impact offensive coaching changes will have on an offense.
A Broad View
To kick off the process, below is a general look at six years of pass/run rates. The rates are split by both quarter and the scoreboard.
|Trailing By…||Ahead By…|
It doesn’t take a stat nerd to catch the obvious trends here. First of all, we see that – for the most part- the pass rates get smaller as we move from the left to the right. This makes perfect sense as teams are more likely to run the ball while ahead and vice versa. We also see some trends from quarter to quarter. Teams are very run-heavy out of the gate and falling behind doesn’t change their game script very much. As we get later in the game, however, we see a healthy correlation between the scoreboard and pass/run rate.
Interestingly, the second quarter is the pass-heaviest frame. The primary reason for this is both team’s attempt to snatch up a few points before halftime. Clubs usually treat the final two minutes of the first half like they’re down by a score in the final minutes of the game. This means quite a bit of passing.
Next up, I removed quarter as a variable and examined league-wide pass/run ratios based only on the scoreboard. I included deficits/leads between one and 32 points and ties. Huge deficits and leads were removed because of the small sample of plays. We were still left with a sample of 200,177 plays, which is 98.5 percent of the entire pot. The results:
Our graph shows some serious non-linear correlation between the scoreboard and a team’s play-calling. The r-squared of .9689 tells us that nearly 97 percent of a team’s pass/run rate can be explained by how many points they’re ahead/behind on the scoreboard. That, my friends, is a massive number.
Of course, we can dig deeper.
Quarter for your thoughts?
Earlier we learned that splitting this data by quarter will give us a more-accurate representation of a team’s offensive philosophy.
My next – and most important – task was to generate formulas that will determine expected pass/run rate based on both the score and quarter.
Note that my sample sizes for these regressions were obviously much lower since I’m cutting the game in four. Also, the score is always significantly closer in the first quarter than it is in the fourth, which means our data will become more reliable as the game progresses.
Right off the bat we see that the score has almost zero impact on play-calling in the first quarter. Coaches generally will not panic if they fall behind early, instead sticking to their gameplan for the first couple series. Consider that the game is tied during 51 percent of plays run in the first quarter. Another 43 percent are run when the difference is no more than one possession (eight points or less).
Earlier I noted that the second quarter is the pass-heaviest frame because of an explosion of passing by both teams near and after the two-minute warning. We see that here. Additionally, we start to see some relationship between play-calling and the scoreboard.
The most popular scoreboard differential during the second quarter is seven points (18 percent). The gap is three points 14 percent of the time, and the game is tied on 13 percent of all plays.
The second half is when we really start to see coaches dump their game script and start calling plays based on the scoreboard. An r-squared of 0.926 makes it rather easy to predict play-calling in the third quarter.
Only seven percent of plays are run with the score tied in the third quarter. Although that may seem discouraging to those of us who enjoy a competitive game, note that 51 percent of plays run in the third quarter are done so during a one-possession game. The differential is more than two possessions only 21 percent of the time.
Whoa! The first three quarters supplied us with some fairly-linear data. The fourth quarter is taking us on a bit of a roller coaster. As bizarre as this chart may seem, it makes perfect sense. The right side of the graph shows the obvious: as a team’s lead increases, so does its propensity for calling a run. The left side is a little trickier in that teams pass a lot when trailing, but will essentially give up on the game and roll with a pass-balanced offense when down by three-plus possessions. Nonetheless, we see a healthy 0.985 r-squared, which makes fourth-quarter play-calling projections very easy.
If you’re wondering why the NFL gets such great ratings, check this stat out: 48 percent of all plays run in the fourth quarter come in either a tie or one-possession game. An additional 26 percent are run during two-possession games. Only 10 percent are run when the differential is more than three possessions.
Next up, we’ll get into the fun stuff. By applying the formulas created from the above graphs to each qualified play from the last six years, we can better understand each team’s offensive philosophy.
Here are the results from 2013 (Table updated 06/26/14):
A brief explanation: The first column shows how often each team actually called pass last season. Note that, as always, it’s dropbacks vs. designed runs. In other words, scrambles are counted as called passes. The second column shows the team’s expected pass rate based on our earlier research. The difference between the two columns is shown in Column 3. The fourth column simply ranks Column 1. Compare that data to our final column, which shows where each team would’ve ranked on the pass/run spectrum had game situation not been a factor.
If we’ve learned anything about the Cowboys under coach Jason Garrett, it’s that they like to pass the ball…a lot. No team preferred the pass more than Dallas last season. The Cowboys ended up as the league’ No. 4 pass-heaviest offense despite holding a lead on 45 percent of their offensive snaps – sixth-highest in the league. Only Cleveland called pass more often when playing with the lead. New offensive coordinator Scott Linehan operated the league’s pass-heaviest offense during his time in Detroit.
The Jaguars jump off the page because of a massive difference in actual and adjusted rank. In Gus Bradley’s first year as head coach, Jacksonville ran only 22 percent of its offensive plays with a lead – third lowest in the NFL. This led to the league’s No. 6 pass-heaviest offense (64 percent). Considering how often they trailed, however, the numbers suggest the Jaguars actually should’ve passed more. It was expected that Bradley would follow the Seahawks blueprint and it appears that is, in fact, his goal. Assuming they’re more competitive in Year 2, expect a lot more running from Jacksonville.
On the other hand, we have the record-setting Denver offense. The Broncos ended the season with a pass/run rate right at league average. Of course, that was only because they went 13-3 and held a lead on an NFL-high 60 percent of their offensive snaps. Denver actually should’ve been among the run-heaviest clubs in the league, but decided against a heavy reliance on its running game. Denver called pass 55 percent of the time while ahead on the scoreboard – ninth-highest in the league. The Broncos were also the No. 2 pass-heaviest offense when tied (63 percent) and No. 5 pass-heaviest when behind (73 percent).
The Falcons, Colts, Chiefs, Saints, and aforementioned Broncos made no major offensive coaching changes and can be expected to remain as pass-first offenses in 2014.
The Bills, Jets, Raiders, Chargers, and Rams preferred a run-heavy attack last season and are expected to bring the same philosophy to the table this season.
There’s plenty more I intend to do with this data and you can bet it will be utilized as part of the Pro Football Focus Fantasy player projections going forward. The next steps include a look at team-by-team and coach-by-coach trends. That will allow for a strong pass/run projection for each team during the 2014 season. Additionally, I’ll take a look at defensive trends so as to supply the best weekly projections during the season.
Follow Mike Clay on Twitter: @MikeClayNFL