Cumulative Effect: Hitting the QB

| May 19, 2014

cumulative-effectEvery time a quarterback is sacked, it’s a terrible thing for the offense. They lose both a down and yards. Even when the quarterback is hit the chances of success for the offense decrease dramatically. A quarterback’s completion percentage drops to 37.8% when they are hit compared to 63.0% when they aren’t. Similarly, their interception rate jumps up to 4.8% from 2.7%.

While that is the initial effect of a sack or a hit, they also have a lasting effect on the quarterback. It has long been assumed that the more you bring the quarterback to the ground, the worse he will do on his following throws. It makes sense from a psychological perspective as the more a quarterback gets hit/sacked in the pocket, the more they will be concerned about getting hit again and the less they think about when and where to throw the ball.

It also makes sense from a physical perspective. The more a quarterback gets taken down by large defensive linemen, the less likely the quarterback will be able to get up with his throwing arm at 100%. It shouldn’t be much of a surprise that the numbers back up the theory. What might be a surprise is the extent that it effects the quarterback.

Below is a chart of how quarterbacks perform in Accuracy Percentage after a number of hits or sacks within a game. For example, when quarterbacks haven’t been sacked or hit yet in a game, they have an average Accuracy Percentage of 73.0%. The sample for this data is every passer in every regular season and playoff game over the last six years.

After every sack or hit the quarterback takes, their Accuracy Percentage decreases by an average of a half of a percent. While that might not seem like much, there also isn’t much difference between the best and worst quarterbacks in the league. Based on the graph, an above average quarterback after five hits or sacks performs as well as an average quarterback with no hits or sacks. Once that above average quarterback has been sacked or hit 10 times, they play as well as a below average quarterback who hasn’t been sacked or hit.

Sacks and hits have a similar effect on how often a quarterback tosses interceptions. Below is a chart for a quarterback’s interception rate based on the number of sacks and hits. For example, when quarterbacks haven’t been sacked or hit, they throw an interception at 2.5% of their passing attempts.

As expected, the more a quarterback gets sacked or hit, the higher the chance any given pass they throw will be intercepted.

In an ideal world from the offense’s perspective, a quarterback wouldn’t let sacks/hits affect him after the play is over. The best quarterbacks in the league don’t let sacks or hits get to them. Here we’ll use a fairly exclusive definition of ‘elite’ quarterback, just grouping Aaron Rodgers, Drew Brees, Peyton Manning, and Tom Brady together. For the following graph we look at all of the first-round rookie quarterbacks just in their rookie year together as well.

Below we compare those two groups as well as the general quarterback population to see just how different they are in Accuracy Percentage. Due to the smaller sample sizes, we only go up to five hits/sacks rather than 10. Also due to the smaller sample sizes, the data is a little noisier.

The elite quarterbacks don’t seem to be effected by the first few sacks/hits. Eventually their accuracy decreases, but at a slower rate than the typical quarterback. Even after a number of sacks/hits, they are still significantly more accurate than the average quarterback who has taken none.

As for the pool of rookie first-round picks, their accuracy decreases more dramatically. On average a sack/hit decreases their accuracy by a full percentage point, which is twice as much as the average quarterback. The perception is that you can rattle a rookie quarterback by hitting them, and the numbers over the last few years back that up.

Statistical analysis should be used to either help prove our conceptions or help change our views. In this case our preconceived notions hold true. This should give offenses all the more motivation to protect their quarterbacks, and defenders all the more reason to get to them.

 

Follow Nathan on Twitter: @PFF_NateJahnke

  • Larry

    Doesn’t player performance generally decline over the course of a game anyway? Due to fatigue and pain and whatnot? I’m curious how the graph you used compares to one done by elapsed game time.

    • KittyTheBear

      You also have to consider multicollinearity here. It’s conventional wisdom that inferior quarterbacks tend to hold onto the ball longer and are hit/sacked more, all things being equal. So the quality of the QB population would tend to decrease as the number of hits/sacks increase. This would cause your graphs to overstate the effect of the # of QB hits/sacks on accuracy.

      • Colin William Weaver

        Yeah, I was thinking that, but the QBs who’ve been hit a lot are first compared to themselves, before they were hit. So yes, those with lots of hits may be worse QBs than those with fewer hits, but even so, the bad QBs perform even worse after getting hit than they did before getting hit. (And so do good QBs.)

    • Joe

      You also have to worry a little bit more about the amount of noise in the plots. Some errorbars would be nice. It looks like the difference in dropoff, for example, among different quality QBs is not statistically significant. When you account for the noise in the data and accuracy dropoff over the course of a game, I’m actually surprised how *little* affect hits seem to have.

  • Colin William Weaver

    Love this. This is super great. A+. I am wondering about a couple controls though. One would be the number of snaps (or dropbacks, or attempts, or whatever), as mentioned below.

    Another would be the pressure a QB faces on each throw, to isolate that it’s the fact that the QB has been hit in the past that’s reducing his accuracy percentage. Perhaps the pressure a QB sees during a game is not static; after getting pressure 5 or 10 times, the defense has figured out what is working and is subsequently more likely to generate pressure in the future than they were in the past. To fully embrace the complexity, it’s not comparing a QB who hasn’t been hit at all to one whose been hit 10 times, it’s comparing a QB who hasn’t been hit and on this particular snap faces no pressure to a QB who has been hit 10 times and on this particular snap faces no pressure, and comparing a QB who hasn’t been hit and on this particular snap faces an unblocked rusher to a QB who has been hit 10 times and on this particular snap faces an unblocked rusher.

    I would guess that the effect would still be evident, but it would be cool to check.

    • Chris from Cape Cod

      Great article, and nice post, Colin. Permutations as you suggested would get to some pretty complex type of analysis, but in a pass driven league with limited elite corners, I think this is a brilliant area of focus, and well deserving of the research time.

  • Dave

    There is a problem here.

    Bad qbs teams tend to be on bad teams. Bad teams tend to be farther behind as the game goes on or when the sack When a QB is farther behind he will throw riskier passes which will be intercepted more and completed. Its likely this is contributing to the reduced accuracy/higher int%. It would be nice to know the average game state at the time of sack/hit

  • CMG8462

    How was the strength of opponents defense controlled for? You play an awful defense who can’t touch your QB you get to throw a lot of accurate passes in the 0-2 sack buckets. If you are playing an elite defense you are getting sacked early and often and the # of throws in the early buckets are limited, which doesn’t weigh down the accuracy percentage in those buckets.

    • bobrulz

      Does that matter though? This is purely by number of hits, and since better defenses will usually hit quarterbacks more, that is pretty much already controlled for, right?

  • Ridgelake

    Correlation or causation?

  • NoFans NoFollowers

    Interesting, given the back drop of the conventional wisdom, “don’t let your QB be a runner or he will get killed ala RG III”. Apparently it really isn’t any safer in the pocket.

    • eYeDEF

      Well, from a pure standpoint of physics it is. F=MA. Force equals mass times acceleration. A QB running at full speed colliding with a defender running at full speed creates a more forceful impact than a defender running at less than full speed colliding with a QB standing idle. On average, the less forceful impact in the latter instance will result in less severe injuries than the former.