At the beginning of the season, I made a few predictions based on a numerical analysis of pre-2023 data. As promised, I will re-visit these predictions to see how I did.
Prediction 1: The Dallas Defense will regress.
Specifically, I said that Dallas would be “outside the top 10 in Points per Drive and Drive Success Rate.”
Well, I was half right. By Drive Success Rate, Dallas dropped from the 5th best defense in 2022 to 16th in 2023 (68.1% to 70.8% DSR allowed), which means they let opponents move the chains a lot more often against them this year.
However, in terms of defensive Points per Drive, they only dropped 2 spots from 3rd best to 5th. So, I was definitely wrong on that one. Despite allowing more first downs, Dallas kept their opponents from scoring.
This outcome of a high ranking defensive PPD with just average defensive DSR is quite rare. Since 2000 of the 120 defenses that have ranked in the top 5 by PPD, only 3 teams have ranked worse than 15th in DSR (2023 DAL, 2007 NE, 2016 KC). So, what is going on here?
My prediction was based on three beliefs about the 2023 season:
#1 Dallas will face harder opponent offenses - I was right about this. In 2022, Dallas faced offenses that cumulatively ranked 10th in PPD and 14th in DSR (1), but in 2023 their opponents ranked 4th in PPD and 2nd in DSR (Bills, Miami, Detroit, San Fran, Philly twice! Yikes).
#2 Dallas will have fewer takeaways - I was right about this too. Regression to the mean almost guaranteed it. In 2022, Dallas ranked #1 in takeaways with 33. That number fell to 26 this year, putting them in a 4-way tie for 12th place.
#3 Dallas will not be good at stopping drives without turnovers - I hit the trifecta, as I was right about this one as well. On non-turnover drives, the Dallas defense ranked 20th in Drive Success Rate and 20th in yards per drive (even worse than 2022).
So, if I was right about all my assumptions, then how were the Cowboys able to limit scoring? Well, unbeknownst to me, defenses are comprised of more than 3 variables. Who woulda thunk it? On paper, I should have been right: per drive they gave up more yards and more first downs. However, those measures are not 100% correlated with points.
Part of the explanation may be that their opponents had much worse starting field position in 2023 (8th longest fields vs 24th in 2022). This required offenses to earn more yards/first downs to get into scoring position, making points that much harder to come by.
Whatever the reasons, their bend-don’t-break defense got bendier (DSR), but not breakier (PPD). I don’t trust those types of defenses as they can easily fail and you might end up in a situation where oh, I don’t know . . . you give up 48 points to Green Bay.
Prediction 2: Jimmy Garoppolo will be efficient in Las Vegas.
My actual prediction was that “. . . he will finish top 10 in both Yards per Attempt and First Down Conversion rate next year.”
Again, I was only half right. Jimmy G only got to play 8 games, but when compared against other QBs (2), he ranked 9th in First Down Conversion Rate, but missed the top 10 in Yards per Attempt ending up 13th. His 7.2 YPA is pretty good and a lot better than Aidan O’Connell’s 6.5 YPA (26th), but it is still a miss by me.
I really didn’t base my forecast on anything other than Garoppolo’s career. When he was the starting QB for the Niners, he was 1st in Yards per Attempt and 2nd in First Down Conversion %. No doubt, a lot of that was due to the Shanahan system and San Fran receivers, but that doesn’t magically make a bad QB put up some of the best efficiency numbers in the league. Garoppolo has skills (quick release, accuracy) that make him good at moving the chains.
However, I specifically did not include his EPA efficiency in my forecast. That was wise. The reason why was because I wasn’t sure how he was going to be used in the Raiders’ offense and it was probable that they would play from behind a lot, which has the potential to kill EPA.
Garoppolo’s strengths are throwing quick, short, on-target passes. He was never good at throwing downfield and he always had a turnover problem. So, of course, McDaniels had him throw long passes on slow developing plays, where he was inaccurate and threw a ton of picks resulting in god-awful EPA per dropback.
But underneath that poor EPA efficiency, he was still pretty good at moving the chains. I’ll never understand head coaches who try to fit square pegs in round holes (Sean Payton, this includes you).
Prediction 3: Justin Herbert will not be a top 10 QB.
Here is what I said: “I can see him being better than last year, but I don’t see a return to top 5 status. I’m predicting a Yards per Attempt and overall EPA efficiency outside the top 10.”
I got this one 100% right.
Herbert had a bad 2022, yet many were still predicting a bounce-back year in 2023. Not me. Injury kept him from finishing the season, but based on the 13 games he played, he was better than 2022, but still ended up 15th in EPA efficiency and only 22nd in Yards per Attempt.
There were 3 primary reasons for my forecast.
#1 He’s not finding receivers and so will check down a lot - I was right about this, but maybe for the wrong reasons.
In 2021 and 2022, he took a longer than average time to throw the ball, yet he made shorter than average attempts. However, in 2023, while his completions were still short (22nd), he had a quicker ttt (19th). He also had a 14th ranked pressure rate, which relative to his TTT, indicates he was throwing quicker to avoid pressure. So this year, maybe it wasn’t that he couldn’t find receivers, but rather that he didn’t have enough time to. Regardless, he threw short.
#2 His O-Line won’t give him great protection - I was right about this too. In 2023, his protection was a little better than the previous year, but still below average (18th Pass Block Win Rate). In 2021, Herbert’s 5th ranked EPA efficiency was partly due to him rarely being pressured (28th). In 2022 that protection went away (23rd PBWR) and his EPA/d dropped to 20th. 2023 was basically a repeat of 2022.
#3 He will face tougher passing defenses - Yep, that happened. In 2022, he faced defenses that ranked 28th in defensive passing EPA efficiency (when not playing the Chargers), so it was almost a given that 2023 defenses would have to be tougher. This season, the Charger’s faced opponents that ranked 6th in preventing passing EPA (per dropback).
Overall, I’m not sold on Herbert as a top tier QB, but he only has 4 years in the league, so he has plenty of time to change my mind . . . you know, in case he was worried about that.
BONUS ROUND! Jalen Hurts
At the close of my prediction article, I made a passing comment that I thought Jalen Hurts would take a step back this year. Here is why I thought that and why I think I was right (even though I didn’t make any specific prediction).
The above is a dashboard of Hurts’ ridiculously good production in 2022 which includes:
- 3rd highest YPA and 4th highest Net Yards per Attempt
- 5th most accurate passing (cpoe)
- 9th best First Down Conversion rate
- 11th best TD Rate
- 3rd lowest Turnover rate
However, hidden in that monument to greatness, are some concerning numbers:
#1 He faced the 25th easiest pass defenses (opd) - Since opponents change every year, the strength of opponent tends to regress towards the mean. Without even looking at a schedule, I knew the pass defenses in 2023 would be tougher.
#2 He abandons too many plays - He will scramble even when there is no pressure. In 2022, he had the 4th highest scramble rate, yet only faced the 23rd ranked pressure.
Even when he is pressured, its his own fault. In the 2-year span of 2020 - 2021, he had the 2nd highest scramble rate and the 6th highest pressure rate. However, he also had the highest Time to Throw of any QB, averaging 3.1 seconds per attempt behind an O-line that gave good protection (Pass Block Win Rate: 11th in 2020, 3rd in 2021).
In other words, even when he is protected, he likes to hold the ball far too long looking for the deep shot, which invites pressure and then he tries to run out of it. QBs like that usually face a lot of pressure (Justin Fields, Russell Wilson). So, given that I thought Hurts would face tougher defenses in 2023, it follows that I thought he would be pressured a lot more too.
#3 His Passing Success Rate ranked far lower than EPA efficiency - He was around the 90th percentile of EPA efficiency value, but only the 60th percentile in success rate. This means that the volume of his successful plays was barely above average, but the value of those plays was far above average (big positives, small negatives). That big of a discrepancy is uncommon and it is rarely sustained over time. What usually happens is a regression of the value per play rank towards the success rate rank.
So, to summarize, I would have predicted that in 2023, Hurts would face harder defenses, with more pressure, resulting in bigger mistakes/less valuable plays. And I would have been right.
This year, he faced the 12th hardest passing defenses with the 6th highest pressure rate, invited by the 3rd highest Time to Throw and he reacted with the 3rd highest scramble rate. His Pass Success Rate stayed about the same (16th), while the value of his dropbacks all regressed towards average:
- His turnover rate jumped from 30th to 12th.
- His Net Yards per Dropback dropped from 4th to 15th.
- His first down conversion rate fell from 9th to 15th and his TD rate from 11th to 18th.
Now, had I actually made specific predictions about Hurts, one of them probably would have been that his EPA efficiency would end up outside the top 10 and I would have been wrong. He finished 2023 with a +0.143 EPA/d, which was 0.001 ahead of Matthew Stafford, allowing Hurts to hold on to 10th place . . . barely.
(1) Opponent defensive performance was calculated by excluded games against Dallas.
(2) QBs were ranked by season dropback volume and the top 31 selected as a comparison group