To give some context into how offenses tend to regress, I prepared some data from the last four NFL seasons on changes in scoring rates overall and specifically within the red zone. First, overall touchdown rates:
As one can see, offenses that are woeful parleying drives into touchdowns tend to perform better the following year. Similarly, offenses that are run extremely hot tend to fall off the next season. The year over year correlation of past touchdown rate to present touchdown rate is just 0.076 (past scoring rates explain 7.6 percent of the variance in future touchdown rate). The 2018 Chiefs and Saints are perfect examples of the latter phenomenon. The 2018 Chiefs scored touchdowns on almost 48 percent of their drives, the 2018 Saints did so on about 46.5 percent of drives. These were by far the highest rates in the sample, rivaled only by the 2019 Ravens at about 43 percent (whom I will talk about later). The following seasons, these rates declined by about 12 percentage points for New Orleans and 11 percentage points for the Chiefs. They were still great offenses; the Chiefs scored touchdowns on 36.7 percent of their drives in 2019 and the Saints at 34.4 percent despite missing Drew Brees for a chunk of the season. The biggest year to year riser in the sample were the 2018 Chiefs, their first season with Patrick Mahomes as the starting quarterback. From 2017 (with Alex Smith) to 2018, the Chiefs touchdown rate increased by a shade over 23 percentage points, about 33 percent more than the next closest team. The Chiefs in 2020 pose an interesting dilemma when forecasting their offensive potency. They saw one of the highest regression totals in 2019 after they blitzed the league in 2018 and usually you bet that after massive negative regression there will be some positive regression the following year. Yet their 2019 scoring rate was still among the highest in my sample. Subjectively, I think they will not be as proficient in scoring touchdowns as they were in 2019 (so more negative regression) but I think given their quarterback they will continue to post scoring rates somewhere between 30 and 35 percent which would be at or near the top of the league. The 2018 Colts massive jump can be attributed to Andrew Luck missing all of 2017. These examples of the 2018 Chiefs and Colts are not extremely useful in trying to find teams in 2020 that can improve on offense given the unusual circumstances around the quarterback situations.
For some insight into how bad teams' scoring rates improve the following year, let me point out more applicable examples. Houston last year saw 10.3 percent jump in touchdowns per drive. Deshaun Watson started all 16 games for the Texans in 2018 and they scored touchdowns on 24.4 percent of their drives, a middling rate. Given what we know about Watson and that offense, we should have expected them to be more prolific on offense. Going into this year, however, we should expect them to take a step in the other direction after such a large improvement from 2018 to 2019. The Bears in 2018 went from a touchdown are of 21.9 percent to 31.9 percent. We can attribute a lot of this to the Bears defense forcing a lot of turnovers and giving Trubisky and co great field position. When the Bears takeaways decreased from 36 in 2018 (led the league) to 19 in 2019 (22nd), the Bears basically lost all their gains from 2018 as they again scored touchdowns on about 22.6 percent of their drives.
Based on their 2019 touchdown rates and their increases in such from 2018, who are the obvious candidates to improve or get worse on offense? This would theoretically lead to a corresponding increase or decrease in wins and could also help pick out some undervalued players in fantasy.
First on the list with a bullet are the Ravens, as I mentioned at the top. The 2019 Ravens had the third best scoring rate in my sample behind the 2018 Chiefs and Saints, whose scoring rates decreased by 11 and 12 percentage points respectively. Lamar did improve substantially year over year and their scoring rate was 30.5 percent in 2018 so I would expect them to settle in the mid 30s just below Kansas City. San Francisco was also a beneficiary of a large bump. The Niners scored 0.335 touchdowns per drive after 0.235 in 2018, partly due to Garoppolo starting a full slate of games. Garoppolo is not on the same plane as quarterbacks like Mahomes, Brees, and Jackson so to expect the Niners to be as effective in 2020 after blitzing the league last year would not be a great bet. I would make the same type of bet against the Vikings this season, especially after trading deep-ball extraordinaire Stefon Diggs. The Raiders, even after adding some pieces to their offense in the draft, saw massive gains in 2019. Carr is merely steady so expecting another year of high-end offense is wishful thinking. Same goes for Ryan Tannehill and the Titans. I highlighted the Titans in my last post (and will talk about them again later), but the spike in scoring rate in conjunction with leading on a very few percentage of their offensive snaps for a playoff team, makes that team a prime regression candidate (especially on offense) for the 2020 season. For teams that should improve, the most obvious answers are the Steelers and Bengals because of the stark contrasts in each team's quarterback situation. Those teams saw the largest year over year drops in scoring rates (19.6 percentage points for Pittsburgh and 14.07 percentage points for Cincinnati). The Steelers without Ben Roethlisberger scored touchdowns on a league low 18.8 percent of their drives, a mark about half that of the Chiefs and less than half of their division rivals in Baltimore. The Bengals were similarly snake-bitten to the tune of a 20.2 percent rate. With Burrow in the fold, one would hope they get something even approaching mediocre quarterback play. One last team I think has the chance to improve a bit is Atlanta. After retooling the offensive line through the draft last season with poor immediate returns, the Falcons only scored on 30.4 percent of their drives, almost a six percentage point difference from the prior year. They still have Matt Ryan, Julio Jones, and Calvin Ridley. With some improvements from their young lineman and the continued steady play of Matt Ryan, I think Atlanta should again be among the league leaders in turning drives into touchdowns.
Shifting gears a bit, another way to spot teams ripe for regression is looking at red zone splits. In a comprehensive look at red zone performance and splits back in 2016, Bill Barnwell of ESPN showed red zone performance is not consistent year over year and that teams cannot consistently outperform their overall offensive efficiency in the red zone year over year. I reproduced the finding and made a series of charts to help contextualize those findings:
The issue is two-fold: first teams that get to the red zone on an especially large or small percentage of their drives move back towards the mean the following year. Teams that turn an abnormally large or small percentage of their red zone trips into touchdowns will also regress towards the mean. The year over year correlation for red zone rate is 0.167 (not terrible relative to most statistics in football) and for touchdowns per red zone trip its just 0.032. I think this finding is intuitive. How often a team gets to the red zone in a year tells us, in general, how well it can move the football which is the fundamental way of measuring how good an offense is (we can get a better idea if we know about starting field position, but that is beyond the scope of this post). We usually use measures like EPA per play or yards per play, which give more insight into how well a team moves the ball compared to red zone rate. The idea behind red zone rate is similar but we are adding a degree of arbitrariness since the red zone magically starts at the opponents 20 yard line (Why not the 19 or 21 yard line? Or the 15? Human beings like round numbers).
Touchdowns per red zone trip takes the noise associated with touchdown rate to another level. As I explained above, year over year touchdown rate has an R-squared of 0.076 (which is weak). Touchdowns per red zone trip takes the noisiness of touchdown rate and strips out any semblance of signal because because we only consider 20 percent of the entire field. Given the R-squared of 0.032, you would be safer assuming all of the offenses across the league will score at similar rates in the red zone than trying to rank the teams ahead of time.
Now we have established that red zone splits are noisy. Once teams get to the red zone, how well they turn those trips into touchdowns carries no signal. How can we apply this to prognosticating the 2020 season? First, teams that relied heavily on scoring in the red zone in 2019 should be worried about their offensive production in 2020. Great offenses can sustain long drives, but great offenses also can rip off chunk plays and the occasional long touchdown. Here is how all the teams in the NFL stacked up in plays per drive and plays per touchdown drive:
Great offenses like Kansas City, Baltimore, and Dallas can be found at the top. They will tend to have longer drives because they are more difficult for opposing defenses to stop. But the top of the leader-board is also littered with middling to bad offenses like the Raiders, Colts, and Bengals. So even if you run a lot of plays per drive, that might mean you are barely moving football on each play.
Some great offenses string together plays on scoring drives while others relied on making splash plays. Kansas City and Baltimore are near the top here, but other great offenses like those in New Orleans, San Francisco, and Seattle are towards the bottom (worth noting the Raiders selection of Henry Ruggs. They had very few splash plays and had to string together a lot of successful series to score points. Selecting a player with Ruggs' skill-set definitely addresses this, though choosing a player solely to fit a need is a dicey proposition). This is all to drive home the point that scoring comes in all shapes and sizes and is subject to a lot of variance because scoring plays make up a small portion of total plays. An offense that required a ridiculous percentage of red zone touchdowns should be expected to struggle more the following year since that might be an indication they do not have many talented skill position players. It could also be a function of how teams call plays (the Ravens run the ball a ton with Lamar Jackson, which yields a lower explosive play rate than throwing the ball) or regression in the explosive play department (applicable to Tyreek Hill in Kansas City).
With all of that out of the way here is how often teams got into the red zone in 2019 and the corresponding touchdown rates once in the red zone:
No team got to the red zone more than Baltimore last season and once they got there they scored touchdowns at the fourth highest rate. This, in addition to their crazy spike in total touchdown rate, is why I expect Baltimore to disappoint, relative to sky-high expectations (I still expect them to squeeze out the division, but it will be close). I am concerned about the Titans. They had one of the highest overall scoring rates in the league, yet barely got deep into opponent territory. They also scored touchdowns in the red zone at by far the highest rate in the league and the second highest in the entire sample. Their 74.3 percent rate was a hair behind the 75.3 percent rate from the 2018 Steelers. Without Roethlisberger this fell a whopping 36.5 percentage points in 2019. The next closest team in the sample to the Titans was the 2018 Chiefs at 70.7 percent which fell back 11 percentage points in 2019. The Titans relied heavily on explosive plays and converting red zone trips to touchdowns. Neither of these are sustainable. Houston was a less extreme version of the Titans, so I would expect them to take a step back next year. The difference between those two squads is I trust the quarterback in Houston much more than the one in Tennessee. San Francisco might be able to offset some of the oncoming touchdown regression with more favorable red zone splits. Again, the Bengals and Steelers should experience massive bounce-backs. Given the talents at receiver I would expect Dallas to be at least league average in red zone efficiency, which would go a long way in avoiding those one score games that tanked their playoff hopes last year. I would say the Chargers and Bucs are prime candidates for positive and negative regression respectively, but given the changes at quarterbacks I think that is up in the air.
I just went through a lot of data, maybe so much so that it is a bit overwhelming to you, the reader. If that is the case, I would ask you to remember that extreme rate stats and splits are bound to regress season to season given the small samples we deal with in football. Use some of the charts I provided to guide you into finding teams that should get better and worse on offense without considering the context of personnel changes. From there, I would apply more nuance to each individual situation. Doing all of that should prepare you for the 2020 season and prognosticate how offenses will look once they step on the field.
No comments:
Post a Comment