So far, I’ve looked at the basics of offensive identity in 2018 (plays from scrimmage; how many fantasy points (FP) those plays generated; how many of those plays were “wasted” from a fantasy perspective; and the relationships between wins and plays and FP) and the running game (including the correlations between running plays and winning, RB carries and FP, and RB1 “share” and fantasy performance). This week, I’ll dig into the passing games of the 2018 offenses.
Let’s review the run-pass mix of last year’s offenses:
The average team ran on 41% of their plays from scrimmage (sacks count as passing plays) with a standard deviation (SD) of plus or minus 4.6%1. Teams in green had at least one SD more “share” of rushing plays than average; those in red were at least one SD below average. In other words, the teams with green bars were “running teams” and those with red bars were “passing” teams.
The next chart combines the Plays from Scrimmage data I showed you last time with the rushing/passing plays share from above. Naturally, there is a great deal of overlap between the teams with a pass-heavy mix of plays and those that had a large number of passing plays:
But not entirely. Not all the “red” teams in % Rushing Plays are “green” in Total Pass Plays. MIN and NYG drop out because their overall number of plays were below average, so although their passing share was high, their pass play total was not. IND climbs into the green because they had a well above average number of total plays multiplied by an above average but not extreme share of passing.
Similarly, although BAL ran on a greater percentage of plays than any team except SEA, the Ravens had so many total plays that they were about average in total passes. MIA was close to average in run-pass mix, but their offense was so slow/inefficient that they were near the bottom in total pass plays.
The next chart shows how efficient teams were (for fantasy) in throwing the ball, by comparing passing plays to the total fantasy points (FP) each team accumulated passing (25 yds = 1 FP, TD = 4 FP).
The chart shows the number of total pass plays on the x-axis, with each labeled increment representing one standard deviation from average (593). The y-axis depicts the total passing FP: the average team scored 259. Again, each increment is one SD from average.
The dotted blue line represents the expected number of FP scored by a team, given a certain number of plays. The equation of the regression line is: Total Passing FP = 0.5625 * Number of passing plays – 75.081.
Teams above the dotted line scored more passing FP than would be expected by the number of passing plays they had: these were relatively efficient passing offenses. Those below the line were inefficient, scoring fewer points than they would be expected to.
The comparison between PIT and GB is telling. They were the top two teams in total passes (713 and 693 plays). But PIT was very efficient and GB was not: 340 FP vs. 270.
I want to note that for two straight years, TB was well above average in efficiency. Whatever Dirk Koetter’s failing as head coach and Jameis Winston/Ryan Fitzpatrick as QBs, they had an efficient passing game (for fantasy) and it will be hard for Bruce Arians to improve that aspect of the team.
Not surprisingly, KC’s passing offense was almost off the charts at producing fantasy numbers, with a total number of passing plays not much above average.
Of the teams in the “black,” everyone except one either changed OCs, drafted a new QB, or both. OAK is stuck with Jon Gruden (technically Greg Olsen is the OC) but it is a little surprising that they don’t cut bait on Derek Carr.
In the red part of the chart, TEN, BUF, and the Jets all repeated in that area. Rookie QBs get the blame in two cases and I guess another injury to Marcus Mariota is the explanation in the 3rd instance. Rookie QBs also are part of the reason ARI and BAL are down there; WAS and CIN are other injury cases. JAX finally moved on from the Blake Bortles experience. The outlier here is DAL, who are probably about to hand Dak Prescott a big contract. Jerry Jones is apparently convinced the problem was either due to the WRs or the former OC, and “fantasy points” is not the best measure of an NFL QB. The Cowboys were only 19th in RUSHING TDs, so it’s not like the running game was vulturing Prescott’s TD chances. It’s possible that in a couple of years we’ll see Prescott as the new Derek Carr, with a relatively big contract and no team success to show for it.
In 2017, the Top 2 teams in passing FP went to the Super Bowl. Last year, it was #6 and #9. Again, both were efficient passing teams. Of course, PIT and BAL are evidence that passing (fantasy) efficiency is not the only path to the playoffs. None of the teams with above average passing FP replaced their QB (although SF was on the line on both ends of that statement since Nick Mullens and C.J. Beathard threw more passes than Jimmy Garoppolo, who should return to the starting role this year).
The R-squared number (0.3063) is an indicator that the number of total pass plays in itself isn’t overwhelmingly important. It tells us that only about 31%of passing FP can be explained by passing volume. QB ability and a host of other factors are much more important (for example, offensive line and sack prevention: the Panthers scored far more FP on fewer passing plays than the Jaguars even with Cam Newton’s bum shoulder in part due to only allowing 32 sacks vs. 52).
Now let’s look at how teams distributed their passes to each position group and how efficient they were doing that, starting with WRs:
On average, teams targeted their WRs 316 times). PIT far led the league in WR targets; the Steelers and TB were the only teams more than two standard deviations above average. ATL, GB, and MIN were all one SD above average. SEA, SF, TEN, and WAS were all 1+ SD below average in WR targets. The only commonality may be that three of the four had injuries to their starting QBs. Generally, these extremes correspond to teams that threw a lot or a little.
Sometimes the share of targets that WRs get makes a difference:
For example, IND was among the leaders in Plays from Scrimmage and Total Pass Plays, normally that would mean they’d be a leader in WR targets. But the Colts had a very low WR Target Share, so their WRs did not accumulate a large raw number of targets.
The Eagles only gave their WRs 47% of the team’s targets. They and the five other teams with low WR “share” not coincidentally were all in the top six in TE share. KC and IND will both able to still have highly ranked fantasy WRs (Tyreek Hill and T.Y. Hilton), but in general, it’s tough to have productive fantasy WRs without a lot of volume.
Here’s how efficient each team’s WRs were in generating FP with those targets:
This chart uses total WR targets (not share) to compare to FP. The format should be familiar from previous charts, with one change. I’ve added a gray band around the regression line. Teams (WRs) within that shaded area scored about the number of FP we’d expect based on their targets. The equation in blue says that last year, the FP scored by a team’s WR corps = 1.9368 * their targets – 44.404. And the R-squared number of 0.6799 tells us that raw target numbers explain two-thirds (68%) of WR fantasy scoring.
Here’s where I stress how important targets are to identifying WRs who will score FP.
Notice that SEA was very efficient when throwing to its WRS, although the target total was so low Tyler Lockett was just WR26. TEN and WAS had QB issues, but I’d say their WRs weren’t very good – and deserved their low target totals. But BAL stands out as throwing too much to an inefficient WR corps.
PIT, as efficient as its passing game was, and as much as it targeted its WRs, probably needs to use them a little less. And NE, NO, and LAC probably should use their WRs more (I’d add KC to that list if Tyreek Hill plays but I doubt he will). Interesting coincidence that all three have old QBs
As a fantasy owner, you’d like to have both efficiency AND volume. But you can live with some degree of low volume if the team is efficient in producing fantasy points (especially if the production is concentrated on one or two WRs – say, the Chiefs with Tyreek Hill or Saints with Thomas). You can also live with inefficiency if there is enough volume (PIT, both Antonio Brown and JuJu Smith-Schuster). But you really hate to own WRs with the worst of both worlds (basically, all the red diamonds teams).
Next up: how teams used their top WR.
Note the definition of “WR1.” Amari Cooper didn’t qualify as a WR1 on either team he played for by this definition, although he was clearly the #1 in Dallas after he got there and would have in OAK if he hadn’t been hurt for a couple of games. Teams like the Giants may have misleadingly low WR1 targets: Odell Beckham Jr. was on pace for 165 targets when he got hurt.
Again, it’s targets that matter for WRs. The 6 of the 7 WRs with the “green” bars ranked in the Top 7 in FP per game. Tyreek Hill snuck in there at #5 for obvious reasons. Although not a WR1, Smith-Schuster was 4th in total targets and WR8 for fantasy. The big outlier was Jarvis Landry – his 149 targets tied Stefon Diggs (another non-WR1) for 7th overall but Landry was WR27.
Target share can matter for predicting future production. For example, if TEN throws more in 2019 and Corey Davis keeps that 45% share, he could breakout (at last). If Sammy Watkins gets Tyreek Hill’s share (and stays healthy, a topic I address in another article this week), he could finally be a stud WR.
It doesn’t always work out. Last year I wrote, “A.J. Green is a leader in target share and his #12 rank in FP/G is probably his floor.” Then he got hurt, torpedoing that thought- although he was a not terrible WR15 in FP/G.
For the 3rd year in a row, the Chiefs’ WR1 was one of the most efficient producers when comparing actual to expected FP. As good as Patrick Mahomes is, he didn’t make Tyreek Hill, and Hill’s potential absence could be a red flag for this offense.
You can see why the Browns wanted Beckham: Landry’s inefficiency stands out dramatically. Some of that could be the slowness in going to Baker Mayfield. It could also be the team misusing Landry – his career catch rate was over 70% in Miami and just 54.4% last year. He didn’t forget how to catch the ball; I think it was the QB or the routes Landry was given.
As with WRs overall, the simple linear regression illustrated in the graph above indicates a lot of WR1 fantasy production can be explained by target volume (R^2 = 0.8825). You want WR1s who get a lot of volume.
I’ll address RBs and TEs in the last part of this series.
1Rounding in the y-axis labels accounts for the seemingly uneven increments in the chart; they are all 4.6% apart, despite the labels that say 4 or 5 percentage points depending on the segment.
