I have, on many occasions, trumpeted my view that you can expect a QB to perform in the future, similar to how he has in the past. While I find that to be about the most non-controversial statement ever, many people bristle at it and have countered that without the context of surrounding talent, scheme, defenses played, etc., the numbers are meaningless. Fortunately, for those folks, they are only mostly wrong.
In the short term, the members of the church of context have a point: a QB can find himself in a bad situation and that will definitely impact his overall performance. However as time goes on, the variance of those contextual variables gets smaller. For example, a bad O-line can be a significant variable in a single year, but over 10+ years, the average talent of a team’s O-line tends to be . . . well, average. So, over time, the sole common variable and the much larger driver of QB performance is the QB himself. In other words, a QB’s career efficiency will tell you, generally, how good he is regardless of context.
In 2020, Carson Wentz was on a bad team and he had arguably the worst QB performance that year. Looking at context told you that he was likely to be much better in 2021, but his career numbers should have also told you that he was probably just going to be average. Folks dreaming of a return to his 2017 glory were just that . . . dreaming. The hard, but obvious truth is that a QB's numbers are far from meaningless: they tell you precisely who he is.
That is why when talking about “the numbers” it is critical to take a comprehensive view and not just pick a few stats in a recent season and hand-waive them away because of reasons. Instead, look at a QB’s career and if his recent performance deviates from that, then dive into the numbers to see what is driving the difference and that will give you a good idea if that will continue or not.
I did that for Brissett, Rivers, and Wentz and wasn’t wrong. So, I’ll do that for Matt Ryan too.
The first step is to look at Ryan’s career efficiency. The following graph shows his EPA per drop-back in the form of weekly z-scores. It compares his numbers to the 31 other starting QBs for that week (grey dots) and converts those values to the number of standard deviations away from the mean. To smooth the data, I have used 16 game rolling averages, so each dot represents a season’s worth of trailing games (1).
The black line in the middle is the league average and Ryan has spent most of his career above that line (until recently). The green line represents 2 standard deviations above the mean and players at that level are easily the top 3 or higher efficient QBs in the league. Ryan has brushed against that top 3 line occasionally, but he didn’t stay there long. This career efficiency track is comparable to Russell Wilson or Big Ben.
Plotting Ryan’s overall career efficiency(2) on a normal curve shows where he is relative to some comparison QBs.
Matty Ice falls between Stafford and Wilson and that translates to around the 79th percentile of QBs. So, pretty freakin’ good.
Also, let’s just pause to recognize the ridiculous greatness of Peyton Manning sitting at +2.4 standard deviations, which puts him at the 99.2 percentile level in career efficiency.
Of course, the elephant in the room is Ryan’s clear downward trend since his 2017 peak. In the last 2-to-3 seasons, he has been at best, an average QB. Last season his 47.0% passing success rate ranked 17th and his 0.035 EPA efficiency was 22nd. That’s not good. Those numbers are no better than 2021 Wentz (46.0%, 0.077).
Is the main factor here age? Maybe, and if so, then pretty much everything I write from here on out is crap and he will be horrible in Indy. However, I don’t think that is the case for 2 reasons:
- I’m a Colts fan and I can’t bear yet another year of thinking the season is over before it starts.
- The numbers tell me it’s not age. Well, it’s not entirely age.
Let’s start with some measures that are relevant, but don’t show up in a QB stat sheet. The blue dots with the numbered rankings are for Matt Ryan’s 2021 season and the smaller blue dots without the rankings are his career prior to that.
Opponent Pass Defense (opd) shows that in terms of passing epa efficiency, Ryan faced the 6th hardest defenses last year. If I judge the Colts’ 2022 opponents by their 2021 performance, that signals a significantly easier schedule with an opd that ranks 20th.
Adjusted Rushing Success Rate (arsr) reveals that the Falcons' run game was pretty poor last year ranking 30th in run success. A good run game is a lift to QB efficiency and the Colts finished #2 in aRSR last year.
Both of these stats predict that Ryan’s efficiency will get a boost next year.
Ryan’s 2021 Time to Throw (TTT) was about the same as his career average. By itself, that is pretty meaningless, however, relative to other stats, it is very revealing.
Last year Ryan had the 3rd highest Pressure Rate on drop-backs (pr%). Relative to his avg TTT, this screams poor pass protection. Further evidence of this is his 6.4% sack rate (sk%), which is 1.4% higher than his career average.
I would love to say that Ryan will definitely see less pressure behind the Indy O-line, but the Colts' pass protection in 2021 wasn’t good (20th Pass Block Win Rate, 23rd PFF grade) and 2022 isn’t shaping up to be any better. Still, Atlanta was one of the worst-rated lines in 2021 (26th PBWR, 31st PFF grade) so, I have to think the Colts’ O-line will be at least somewhat better . . . although I’m not 100% sure about that.
Another useful purpose for TTT is comparing it to passing depth.
Historically, Ryan has thrown quicker than average passes for longer than average distance (ttt, adot), but last year his average depth of target dramatically fell 1.1 yards to a 27th ranking while his TTT inched up.
This is a sign that a QB is having trouble finding open receivers. His average depth of completion (11th ay/c) was still above average, so he’s not just settling for check-downs, but he’s not finding his first reads.
In 2021, he targeted his wide receivers 5% less often then his career average and when he did, the average depth of target was 2.7 yards shorter. His #1 receiver was a tight end (Kyle Pitts). In addition, his rate of attempts over 20 yards (20+), fell to 30th in the league, even though he was highly successful on those passes (4th highest avg epa). That all points to a QB not finding open wide receivers downfield.
Perhaps that is due to scheme, good opponents, poor receivers or maybe the pressure just gets home too soon. What it probably is not, however, is that Ryan just suddenly lost the ability to see the field as that is not an age-related thing.
Prior to 2021, Matt Ryan’s 65.6% completion rate had been about league average, but last year it actually increased to 67.3%. However, that is not a very good way to measure QB accuracy.
Shorter passes are easier to complete and Ryan’s increase in cmp% came with a significant decrease in the depth of targets. A good accuracy stat will adjust for passing depth, which is what completion % over expected does (cpoe). Even though Ryan’s cmp% went up, his cpoe went down.
The good news is that while accuracy varies from year to year, it isn’t something that typically falls off with age. The following chart shows season cpoe since 2008 for QBs that were at least 30 years of age (3).
The trendline is basically flat. The tiny increase in accuracy with age is likely a survivor bias in the data, as QBs with good cpoe have longer careers.
The point is that Ryan has historically been a top 10 QB in accuracy and even though that dipped in 2021, he was still above average and there is no age-related reason to think he can’t return to the top 10 again next year.
Ryan’s receivers didn’t gain many yards after the catch last year (27th yac). That’s not simply due to passing depth either as their 25th ranked YAC over expectation (-0.6 yacoe) was far below average too.
Now, QB accuracy is also a big factor in YAC, but I just showed that Ryan’s accuracy was better than average and with that should come at least average YACOE. So, this points to receiver issues.
The Colts aren’t by any stretch of the imagination a receiver-heavy team, but last year, even with Wentz’s horrible accuracy (28th cpoe), the Colts’ receivers managed -0.2 YACOE which is about half a yard better than what Matty Ice experienced. So, even if the Colts don’t improve much at receiver, Ryan should see more completions with longer yards.
Matt Ryan’s efficiency has never been driven by running around and throwing the ball out of the stadium. So, age-related losses in mobility or arm strength aren’t really concerns for me. He has always been a QB to get rid of the ball on time to the open guy underneath and last year he played with an O-line and receivers that didn’t let him do that.
He also faced good opponent defenses without a complimentary run game, so it should be no surprise that his efficiency was down. I think the Colts solve some of those problems (at least partially) and therefore, Ryan sees an efficiency improvement in 2022.
I will stand by my conviction that if you want to know a QB’s future, look to his past. Over his career, Ryan has averaged a +0.16 EPA per drop-back and a trailing 5 year average of +0.13. I think that is a reasonable range to expect for next year, which for context, would have ranked as the 12th-13th most efficient QB in 2021 and fallen between 2020 Rivers (+0.22) and 2021 Wentz (+0.08).
The bad news is that I’m not sure that an improved passing game will necessarily translate to more points. I say that because the 2021 run game was so ridiculously proficient, that it likely will take a step back and my guess is that the passing gains will simply offset that.
Still, that’s not a bad thing as the offense’s 2.45 points per drive(4) was 10th highest last year. If the run game were to maintain similar heights in 2022, then I see about a +0.2 to +0.3 increase in points per drive, which would likely make for a top 5 scoring offense.
1) QB Rolling averages were calculated off of the prior 16 games played or their total games if they had not yet accumulated 16 career games.
2) For career z-scores, I limited each year’s data to only QBs with at least 200 attempts. I then took their epa/d by season and converted that to a z-score for each year. I weighted the yearly z-scores by season drop-backs to get a career weighted average. QBs with less than 1,000 career drop-backs were removed and I re-normalized the scores for the remaining QBs.
3) Includes only the top 32 QBs by passing attempt from each year since 2008.
4) PPD excludes non-scoring drives that end with time expiring and reduces points by opponent safeties and return TDs off of turnovers.