Insert lengthy intro here. I uncovered three traits/trends for the Giants. Now I’m going to write about them.
The Giants’ defense is playing outstanding football and they’re going to get more sacks.
This isn’t really a “trend” per se, but I wanted to point out that the Giants are playing extremely well on defense. They’ve allowed only 6.0 net-YPA through the air and 3.6 YPC on the ground. Grading defenses as a function of expected points allowed, Advanced NFL Stats has them ranked as the fifth-best defense in the NFL. Meanwhile, Football Outsiders, which ranks teams independently of opponents, has them at No. 9.
That’s all the more amazing since the Giants rank dead last in the NFL in sacks with only 14. Sacks are extremely fluky, however, and the G-Men are doing a decent job of reaching the quarterback with 114 pressures on the year, according to Pro Football Focus. Based on that total, they should be closer to 25 sacks, meaning they’ve just been extremely unlucky to acquire 11 less than that.
So here’s my New York Giants defensive sack prediction that I know you’ve all been waiting to see. The Giants, a team that’s generated only 1.4 sacks per game this year, will double that mark over their final six games (for somewhere in the neighborhood of 16 more sacks). It will start with a few this weekend against the Cowboys.
Eli Manning shows play-action on 14.9 percent of his dropbacks.
That number ranks him 24th of 29 qualifying quarterbacks, according to PFF, with Tony Romo of course No. 29 at 10.3 percent. The play-action numbers are so important because the play-type is incredibly underutilized across the NFL.
To give you an idea of how successful teams are on play-action, take a look at some of these numbers. First, yards per attempt and touchdown rates…
As a whole, NFL quarterbacks have averaged 8.62 YPA on play-action passes, compared to just 6.87 YPA on all other passes. That’s a jump of 25.5 percent, which is remarkable. It’s true that play-action passes might be used in more favorable game situations than other passes at times, but the situations aren’t so different that we could expect such a dramatic deviation in YPA.
You can also see that quarterbacks have thrown a touchdown on 5.7 percent of their play-action passes, compared to 4.1 percent of their other passes. That’s an even larger increase of 39.0 percent. Also consider that quarterbacks have thrown 2.45 touchdowns for every pick on play-action passes, but only 1.48 touchdowns for every pick otherwise. That’s a 65.5 percent increase!
I’ve shown again and again that you don’t actually need to be able to run the ball to use play-action since defenders play situations, not past rushing efficiency. So regardless of either team’s rushing woes, it’s a tragedy that they use play-action passes so infrequently.
The Giants can’t properly randomize their plays.
One of my favorite areas of play-calling to study is that on 2nd down because I think it can tell you a lot about an offensive coordinator’s mindset. Specifically, I like to look at 2nd and 10 because it’s a situation in which the offense often threw an incomplete pass on first down.
Most NFL play-callers can’t randomize their play-calling. Humans in general are poor at replicating randomness, usually alternating occurrences much too often. If someone asks you to guess the result of 100 separate coin flips, you probably won’t have a string of five straight heads (or tails), even though that’s likely to occur just by chance.
NFL play-callers are the same way, often calling a run after a failed pass and a pass after a failed run. They think that by “mixing it up” they’re randomizing their play-calling, but that thought process is ironically making their choices very predictable. The Cowboys were actually one of the worst second-down play-calling teams in the NFL prior to hiring analytics guru Ken Kovash, who wrote a paper on the topic and helped fix the problem. He’s since departed for Cleveland.
Well, the Giants, like most teams, can be predictable. Below, I charted the pass rates for the Giants, Cowboys, and NFL as a whole on 2nd-and-9, 2nd-and-10, and 2nd-and-11.
The rate of passes on 2nd-and-10 is lower than that on both 2nd-and-9 and 2nd-and-11 for all three groups. That’s what we’d expect if teams aren’t properly randomizing play-calls, following incomplete first down passes with too many second down runs. In reality, we should see approximately equal pass rates on all three down-and-distances, or perhaps a slightly higher pass rate on 2nd-and-10 than on 2nd-and-9.
This is a small subset of a much larger issue: teams suck at calling plays in an optimal fashion. The Cowboys can take advantage of this by understanding when the Giants are most likely to run or pass; it should be dependent on the down-and-distance and game situation but independent of the previous play-call (run or pass), but it’s not.