Throwing deep and accuracy in the NFL

The farther you throw the ball in the NFL, the lower the chance of a completion. If you average the 2018 and 2019 seasons, passes thrown 5 yards from the LOS were completed 72 percent of the time. Compare that to a completion percentage of 40 percent for passes thrown 30 yards from the LOS. It would stand to reason that quarterbacks who throw more deep passes are going to be less accurate. We shall dig into that in this article.

Depth of throw will be tracked by a stat called average intended air yards, IAY, which is available from NFL NextGen Stats. In 2020 Drew Lock (9.1) was second only to Tom Brady (9.3) in IAY among the 35 full-time starting QBs. Joe Flacco and Jalen Hurts both had a higher value, but neither was their team’s starting QB.

Pro-football-reference.com tracks passing accuracy with a stat that they call OnTarget%, which tracks how often a QB’s throw is on target. Spikes and throwaways are excluded. In terms of ontarget%, Lock was dead last (35th out of 35) at 68.9 in 2020. This was a significant drop for him relative to 2019 when he had a value of 73.0 percent - on about on third of the total dropbacks. Some have looked at Lock’s IAY and his ontarget% and concluded that his inaccuracy resulted from his deep pass percentage in 2020. Was this the case for Lock? Does this trend hold true for other NFL QBs - i.e. is there a negative correlation between ontarget% and IAY?

For Lock the answer is a definite yes. According to NFL nextgen stats, Lock had an IAY of 6.9 in 2019 when he had on ontarget% of 73.0. In 2020 his IAY went up significantly to 9.1 and his ontarget% dropped significantly to 68.9. What about the rest of the starting QBs in the NFL?

For 2020 there was only a correlation of -0.10 between IAY and ontarget% among the 35 starting QBs. That’s almost negligible. What’s interesting is that if you go back to 2019, the correlation is -0.37, still not great (from a statistics perspective), but for sports data that is a fairly strong negative correlation. This begs the question, why the change year over year?

2019 IAY and ontarget%.

2020 IAY and ontarget%.

One possible reason could be that QBs had more time to throw deep routes without being pressured because of the historic drop in offensive holding calls.

If the pass rush can’t get there, QBs have a better chance of making more accurate deep throws.

Another possible explanation is that it was just a good year for deep passes with a significant number of QBs individually improving their deep ball accuracy year over year. A few examples follow:

We are going to avoid the discussion of Cam Newton because he had so few passes in 2019 - 95 dropbacks. Obviously health and change of team/scheme helped him improve so dramatically.

Ryan Fitzpatrick had a large change in accuracy year over year. He went from starting thirteen games and having 542 dropbacks in 2019 to having half of that in 2020. He started seven games and dropped back to pass 281 times. One possible reason for his improvement is the improvement in his offensive line. His sack rate in 2019 was 10th worst (7.4 percent) and his sack rate in 2020 was 14th (5.0 percent). Looking beyond sack rate, his pressure rate in 2019 was 26.7 percent (7th worst of 32) and that dropped to 23.0 percent (15th worst of 35). He was under pressure less often in 2020. That helps accuracy. Just ask Patrick Mahomes how pressure affected his accuracy in the Super Bowl. Fitzpatrick had an IAY value of 9.0 in 2019 and that dropped 7.9 in 2020. So that also could have been a big part of his improvement in accuracy.

Another guy who improved his accuracy greatly from 2019 to 2020 was Matthew Stafford. Stafford was protected really well in 2019 (7th lowest pressure rate - 19.1 percent) and was still near the bottom in terms of accuracy (30th of 32). His line did a similar job of protecting him in 2020 (21.2 percent pressure rate - 14th lowest of 35) so OL protection can’t be the reason why Stafford improved. Maybe Stafford was motivated to play his best since he thought it would be his last year with the Lions? PFR only has accuracy data for the past two seasons. It would be really interesting to see which accuracy number, 70 percent or 77 percent, is more reflective of his other ten years in the league. Stafford led the league in 2019 in IAY with a value of 10.7. His IAY dropped down to 9.0 in 2020. I’m sure that helped improve his accuracy year over year.

If we track Stafford’s IAY value back as far as it goes at NFL Next Gen Stats (to 2016) we find this:

• 2020 - 9.0
• 2019 - 10.7
• 2018 - 7.0
• 2017 - 8.0
• 2016 - 8.0

Stafford appears to be right around the league average in most years and has only been above it in the past two.

Here is the league average in average IAY and the standard deviation by year:

• 2020 - 7.97 plus or minus 1.09
• 2019 - 8.14 plus or minus 1.06
• 2018 - 8.10 plus or minus 1.07
• 2017 - 8.33 plus or minus 1.13
• 2016 - 8.87 plus or minus 0.99

So passer rating and completion percentage are going up year over year and depth of target is going down (at least over the last five years).

Maybe a better way to look at this is to look at the average ontarget% and average IAY for the starting QBs in both years. In 2019 the average was ontarget% was 75.1 percent. In 2020 that went up to 76.8 percent. That might not look that impressive but you should keep it in context. The 35 starting QBs in the league dropped back to pass 16735 times in 2020 (not counting scrambles). So an increase of 1.7 percent means that roughly 300 more throws from starting QBs were on target in 2020 relative to 2019. However, that only works out to about ten more ontarget throws per starting QB. Not counting scrambles there were 19153 dropbacks in 2020. So starting QBs accounted for 87 percent of dropbacks.

The average IAY among the 39 listed QBs in 2019 was 8.14. The average IAY among the 41 listed QBs in 2020 was 7.96. Again this change of 0.18 yards may not seem like much, but it could explain the year over year improvement in league-wide accuracy. So Matthew Stafford in 2019 was more than one standard deviation above the average. In 2020 Matthew Stafford’s value of 9.0 was less than one standard deviation above the average. In other words, he was an outlier in average throw distance in 2019 and not one in 2020.

Now let’s look at some of the guys who got much worse in 2020 relative to 2019 in terms of accuracy. Only two quarterbacks dropped by more than four percent, Drew Lock and Dwayne Haskins.

Dwayne Haskins had an IAY of 8.7 in 2019 and an ontarget% of 74.0. In 2020 his IAY fell all the way to 6.7 (outlier on the low end) and his ontarget% dropped to 69.6. That 69.6 was the lowest in the league among the 35 starting QBs and only Haskins and Lock were below 70%. What might explain Haskins’ drop in accuracy given that his average depth of target was two whole yards shorter? It wasn’t pressure. Haskins was pressured on 31.1 percent of his dropbacks in 2019 and only 22.8 percent in 2020. Maybe it was a change in offense and offensive coordinator. In 2019, the OC for Washington was Kevin O’Connell. In 2020 it was Scott Turner. I’ve been told that changing offensive coordinators is not good for young quarterbacks, but I have yet to verify that statement with data.

The same excuse could be (and has been) for Lock’s worse performance in 2020 relative to 2019. In 2019 Lock had an IAY of 6.9 and an ontarget% of 73.0. In 2020 his IAY climbed all the way to 9.1 which was second only to Tom Brady’s 9.3 among regular starters. His ontarget% fell to 68.9 in 2020 as you might expect from a second year QB who was attempting significantly longer throws on average than he did as a rookie. This, of course, makes me want to see if other QBs from the 2019 draft increased both their depth of target and their accuracy year over year. The other three QBs from the 2019 draft who we haven’t discussed yet that played, all improved their ontarget% from 2019 to 2020. Daniel Jones went improved from 71.1 to 75.2; Kyler Murray from 73.5 to 77.1; and Gardner Minshew from 73.6 to 76.5. Jones and Minshew did it with new offensive coordinators as well.

So how did those three compare in their IAY in 2019 and 2020. Jones went from 8.1 to 7.7. Murray went from 7.1 to 8.0 and Minshew from 7.5 to 8.2. So only Jones was throwing shorter on average. Now, neither Murray or Minshew had as large of a year over year increase in IAY as Lock, but they both improved their accuracy with a larger IAY while Lock had one of the worst drops in accuracy among starting QBs.