Thanks to NFL Next Gen Stats, Pro Football Focus and the nflFastR project for being awesome sources of data for this analysis.
Philip Rivers was a quarterback who got rid of the ball quickly. You can argue about why that is, but you can’t argue if it’s true. Time to Throw is a stat that measures the amount of time between snap and the release of the football on a pass attempt and last year Rivers’ average TTT ranked 29th (4th quickest).
Over his career, Carson Wentz has averaged the 13th longest TTT and in 2020 he ranked 5th longest. So, it didn’t take a PhD in rocket surgery to guess that in 2021 Wentz would take longer to get rid of the ball than Rivers. And surprise, after week 5, Wentz has had the 12th longest average Time to Throw of the 32 QBs with the most attempts.
So what? Well, maybe nothing. Maybe a lot. It just depends.
TTT is not a stat like Yards per Attempt or EPA per drop-back, which are representations of play outcomes. Those stats measure the value of the passing game: the more value the passing game adds, the higher the numbers. So, there are “good” and “bad” measures of YPA or EPA/d.
Conversely, TTT in isolation tells you nothing about the value of the passing game. There is no “good” or “bad” TTT. The following chart is the season average TTT and the Yards per Attempt of the top 32 QBs by attempt from 2016 - 2020.
If you think that looks pretty random, you are correct. There is a slight trend that YPA increases with higher TTT, but the correlation between the two is very low (+0.12), meaning that a highly efficient QB could just as easily have a high TTT as he could a low TTT. So stating the fact that Wentz holds the ball longer than average is not a comment about the value of his play.
However, relative to a measure like passing depth, TTT explains a lot. There is a moderately strong correlation between passing depth (aDOT) and TTT (+0.40), which means the longer the average attempt, the higher the TTT on average.
Basically, if a QB is holding the ball a relatively long time but only attempting short passes, this can be an indicator that something is wrong. Maybe he can’t find open receivers or he is too conservative. Any Colts fan who watched Jacoby Brissett in 2019 should be nodding their head right about now.
Now, even though there is a clear signal in the data, there is still a lot of noise (dots all over the place), and accounting for variables such as play-action, yards to gain, and field position would probably help clean that up, but I don’t have TTT at a play level, so I can’t do that. I’ll just have to make do with what I have and be conscious that this analysis is more of a chainsaw than a scalpel.
Another strong relationship with TTT is pressure. The correlation between TTT and the combined sack + scramble rates of QBs is +0.53.
But this begs the question, does high TTT invite pressure, or does pressure invite high TTT? The answer, of course, is yes.
Both things are true, which makes understanding game-level TTT complex (not impossible, just complex). To simplify it, let’s begin by removing pressure from the equation. Pro Football Focus splits their passing data into “pressure” and “non-pressure” drop-backs. I use quotes there because these are subjective categorizations and I have no idea what their criteria are for each category, but their charting data is pretty good, so I will trust it.
Below are the QB season average depth of targets by TTT for both pressure (left) and non-pressure (right) attempts from 2016 - 2020.
There is a lot to unbundle here.
- First of all, notice that the TTT is greater in general in the pressure graph (avg of 3.3. vs 2.4), which makes sense. Whether pressure causes TTT or the other way around, the TTT should be higher on pressure attempts.
- Also, notice that the aDOT is greater in the pressure graph (avg 10.7 yds vs. 8.0), which is an important result. Even when under pressure, QBs on average get longer passing depths out of higher TTT situations. Regardless of the causation, one comes with the other.
- Lastly, it may not be obvious, but even with these differences in measures across graphs, the correlations between TTT and passing depth from both charts are similar to each other (+0.36, +.039) and they are similar to the overall combined data from which they came (+.040).
This last point is crucial to understand because it is saying that splitting the data by pressure does not provide any more clarity of the relationship of TTT to aDOT, rather it is simply two different views of the same thing.
According to PFF, when not under pressure, Wentz has the 17th longest TTT and the 3rd shortest passing attempts after week 5. So, while his TTT is about average, it still seems high relative to how far he throws, but let’s chart the data to verify (Wentz is the red dots).
You can see that indeed his 2021 measure is above the trend line indicating that he is relatively slow for the average depth thrown. The same was true for his 2020. Again these are on attempts without pressure, so you can’t really blame the O-Line for this.
Based on the linear regression in that chart, I estimate that 2021 Wentz is 0.12 seconds slower than expected. Now, that may seem too small a value to matter, but give the average QB an extra 0.12 seconds and he finds receivers 2.6 yards further down the field per attempt. If he completes 65% of those, then that is 1.7 yards per attempt or about an extra 50 or so passing yards a game on average. It matters.
When looking at the pressure data, Wentz is still a little slow, but closer to the expected passing depth.
So what have we learned?
- TTT by itself means nothing
- TTT relative to passing depth does have meaning
- The presence or absence of pressure does not alter the relationship between TTT and passing depth, so passing depth is not a function of pressure.
And this last bullet point is what many people don’t seem to understand. Carson Wentz has had the 2nd highest pressure drop-back rate of any QB, which naturally shifts his TTT higher. However, that comes with a correspondingly higher aDOT of 9.1 yds vs 6.1 yds on unpressured attempts. So, the amount of pressure he has experienced does not explain the disconnect between his high TTT (12th) and short aDOT (30th). I’m going to repeat that:
The amount of pressure he has experienced does not explain the disconnect between his high TTT and short aDOT.
That does not necessarily mean that Wentz has played poorly. The relationship of TTT to aDOT is like the #10 thing I care about when measuring QB play. It is simply a descriptor that helps add another dimension to passing depth to identify disconnects and potential issues in play that may impact other more important measures of the passing game.
- 2017 Jacoby Brissett had a short passing depth (27th) and a long TTT (5th). Disconnect.
- 2018 Andrew Luck had a short passing depth (23rd) and a short TTT (25th). No disconnect
- 2019 Jacoby Brissett had a short passing depth (21st) and a long TTT (6th). Disconnect
- 2020 Philip Rivers had a short passing depth (25th) and a short TTT (29th). No disconnect
You tell me which years had a problematic passing game.