Here at Holy Schwartz, I like to bring you data analysis that no one else has ever done before. Whether it’s studying arm length data (which I recently updated for PFF) to see if Riley Reiff can cut it in the NFL or comparing Reggie Bush to Jahvid Best on a carry-by-carry basis, having my own blog without deadlines affords me the opportunity to spend more time than I should compiling these unique data sets and searching for some meaning within the numbers. This type of work is simultaneously the most rewarding and the most criticized type of work because it doesn’t always lead to familiar quantities or hard-and-fast conclusions, so please bear with me.
Over the past month or two, I’ve been working on a project to interpret snap count data in unique ways. In August 2012, the NFL announced that it would begin including individual snap count data in its official play-by-play gamebooks. Football Outsiders compiles this data for each team, week, and position and has made it publicly available. The most obvious use for this data would be to provide rate stats (targets per snap for wide receivers, pressures/sacks per snap for defensive linemen). While those quantities are useful, they don’t really tell a drastically different story than what’s already out there – just maybe a more accurate one. My idea is to use the snap counts themselves to provide a completely different type of analysis.
The first thing I can do is make a de facto depth chart for the year. It’s not necessarily who starts all the games or who is first choice when everyone is healthy. Instead, you’re simply sorting by number of snaps at each position. It’s presumably the best, healthiest, most willing to line up in the correct spot (ahem, Titus) guys on average throughout the year. It’s a depth chart that accounts for health.
There’s nothing truly groundbreaking here. It’s all more or less how you’d expect it to shake out if you paid attention to the team last year. The interesting things to note here are the percentages of snaps each guy gets. With some positions (like WR, DB, and DL), there more of a gradual decline in snaps as you go down the depth chart rather than a sharp cutoff, while positions like OL and LB are pretty much just comprised of starters and there aren’t many backup snaps to go around. Using this data, I can also total up snap counts by position. This gives us an idea of formations (or at least personnel-based formations). As you can see from the above tables, on average, the Lions played with 1 QB, 2.45 WR, 0.96 RB, 5.16 OL, 1.42 TE, and 0.01 FB. Similarly on defense, the Lions used 4 DL, 2.6 LB, and 4.39 DB.
Since all of this is done by sorting and a little bit of math, I don’t need to know anything about the Lions roster to put any of these numbers together. More importantly, I don’t need to know anything about any of the other teams in the league to put theirs together. So in doing that, I can see how the Lions stack up formation/personnel-wise with everyone else.
To me, the interesting things on the offensive side are that the Lions aren’t all that heavy on wide receivers when compared to league average, but are very heavy on OL and TE. This doesn’t really fit the narrative of what we expect from the Lions offense – spread the formation, put 5 WRs on the field, and throw the ball down the field. This probably has something to do with Reiff being listed as an offensive lineman and Scheffler being listed as a TE, when in reality, Scheffler played a lot more slot wide receiver and Reiff played a lot more of that 6th O-Lineman/Tight End role. What’s really interesting is on the defensive side of the ball, where the Lions are last in the league in DB usage. There doesn’t seem to be any unusual cross-position confusion that would skew things. The best explanation I can give for this is that the Lions put a premium on getting after the QB (hence the 5th most DL usage in the league), but can’t sacrifice linebackers to get more linemen on the field because their scheme leaves them vulnerable against the run. Couple that with the massive amounts of injuries in the secondary and the Lions might have felt more comfortable with linebackers covering tight ends than guys way down the DB depth chart.
The next thing I can do with these numbers is combine the depth chart and the formation analysis to say something about depth at each position. The idea is that if the Lions use 2.45 wide receivers per play, then in an ideal world, you would have the top 2 guys play 100% of the snaps and the 3rd guy play 45% of the snaps. As soon as you start using your 4th wide receiver (or use your 3rd wide receiver more than 45% of the time), you’re starting to utilize your depth. So I can then figure out the proportion of snaps taken by non-starters, or what I’ll call Depth Utilization Ratio (DUR). And as I said above, I don’t need to know anything about the rosters of other teams to do this, so here’s a chart that lists the DUR for every team at every position and how the Lions rank.
The individual positions have DUR listed as a ratio of non-starter snaps per total snaps for that position. The total offense, total defense, and overall are listed as non-starters per snap (so out of 11 players on the field).
What we see here is very interesting. First of all, the numbers all seem to make sense, which is a good check that it’s describing what I want it to describe. Second of all, it gives you an idea of how the Lions compared to the league averages and just how those averages shake out. You can see the average DUR for RBs are huge compared to all the others, while WR and DL are also pretty high and fullbacks are wayyy low. The fullback one is easy to explain. If you even have a fullback, you pretty much just have the one fullback. Not much rotation going on there. As for running backs, it seems like running back by committee is even more popular than I would have thought. What this is saying is that starting running backs only take about 58% of the snaps on average. Wide receivers and defensive lineman make sense in that they’re constantly moving, so they need to be swapped in and out a lot. Also, formations often range from 1 to 5 wide receivers, the biggest variation of just about any position, while defensive lineman are swapped in and out depending on the situation (running or passing down).
Looking specifically at the Lions, the most shocking thing is their DB depth utilization. It’s the highest in the league (by a huge margin) and over double the league average. Almost 40% of the Lions DB usage was from non-starters. It just really goes to show how much of a revolving door that position was throughout the year. The other top 10 positions are defensive line, running back, and wide receiver. Those were all pretty bad units last year, so it would stand to reason that was either causing or caused by having to go way down the depth chart. The positions that were rock solid were offensive line and linebacker. Both were among the lowest DUR in the league at their positions, which is no surprise because of the health and lack of depth at both positions.
All in all, this seems like a useful indicator of depth usage and a good way to compare across the league. Hopefully, I’ll be able to look into this more in the future to see how this changes with Lions roster changes. It would also be nice to combine this with PFF data and standard team statistics like yards and points to see how DUR correlates with success. Feel free to leave comments on ways to improve this. It should only get better with discussion and tweaking.