The Cowboys called 15 play-action passes against the Giants in Week 12 after averaging 4.1 in their first 10 games. Hopefully that was the result of the team recognizing a weakness and correcting it.
And while the Cowboys certainly deserve some praise for making the change, you have to wonder what took so long. I mean, I was calling for more (and different) play-action types since I started tracking and blogging about the Cowboys while in college in 2010. That was three seasons ago, and the team is just now catching on. There’s something wrong with that picture.
One of the common rebuttals to the implementation of analytics in football is that you need to “turn on the film” or something needs to “pass the eye test.” Stat geeks get criticized for small sample sizes in the NFL, yet you want to reduce the sample even more and trust one man’s opinion in reference to limited film study?
In an e-mail thread from earlier this week, one of my fantasy football cohorts summed it up nicely:
I think one logistical problem that some of the media scouts have is that it's just not possible to watch every play for every prospect. So they end up watching a few games and hope they get a representative sample, which is probably problematic. This is probably a good example of the blind men and the elephant. They're all describing what they're observing, and they might all be part right, but they're describing different parts of the elephant.
We don’t want to describe different parts of the elephant, do we? No, we want to take an approach that captures the entire picture.
Another lackluster criticism of stat analysis is that it can be misleading at times. Maybe that’s true, but it’s hard to argue that stat analysis falls victim to those defects to a greater extent than film study.
When I tell you the Raiders’ first down run rate later in this article, there’s nothing at all misleading about those numbers. We might interpret them in different ways, but go ahead and watch every player from every Raiders game and then tell me the frequency and efficiency with which they call plays on first down. The numbers quickly and accurately capture many aspects of football that aren’t easily discernible through film study.
The key to analytics in the NFL isn’t about certainty; it’s about exposure to optimal conditions. If a defensive coordinator knows that an offense throws the ball on half of their 2nd-and-1 plays, that information is extremely useful. He doesn’t need to know for sure if a run or pass is on the way on a given 2nd-and-1, but continual exposure to the optimal defense given the opponent’s tendencies will result in the greatest advantage over the long-run.
So with that said, let’s take a look at a few Raiders tendencies heading into Week 13.
8.25: The length of Matt McGloin’s hands
So after 500 words on recognizing and adapting to trends, I’m going to start off with the hand size of the Raiders’ quarterback. Makes sense.
Hand size isn’t a tendency, obviously, but I believe it’s the most overlooked trait when assessing quarterbacks. NFL teams value height in a big way, but I’ve found that hand size might be way more important for passers.
We continually see the tallest quarterbacks perform the best, but perhaps not because of their height. Height is strongly correlated with hand size, so if hand size were really a strong predictor of quarterback success, we’d expect it to be reflected in the relationship between height and career stats.
But when you start to analyze some of the short quarterbacks with unusually large hands—Russell Wilson and Drew Brees among them—you see that hand length is an even better predictor than height. Teams are valuing the wrong trait, so they could actually acquire value in the draft by searching for short quarterbacks with large hands.
My guess is that large hands help quarterbacks control the football and throw it accurately. That’s perhaps one reason why McGloin—who has unbelievably small hands—has been horribly inaccurate over his football career. In three seasons at Penn State, he completed fewer than 60 percent of his passes, and his completion rate through three NFL games is just 55.7 percent.
So can this information be translated into actionable advice? I think so. Namely, I’d advise Dallas to play conservative defenses against McGloin. It’s popular to want to put pressure on young passers, but that’s a high-variance strategy that really just increases the probability that they generate quick scores.
The numbers suggest that it’s unlikely McGloin will continually make accurate throws to move Oakland up the field, so make him do just that. Play Cover 2 and Cover 3 to force him into making accurate throw after accurate throw.
43.5: Raiders’ first down pass rate through three quarters
The Raiders run the ball a whole lot on first down. Check it out.
Their first down pass rate barely increases over the course of games, always sitting below the league average. Meanwhile, the Cowboys run the ball on first down way, way too often to start games. They correct that by the second quarter, but they really need to consider scrapping the early first down runs.
88.9: Percentage of pass-rushing snaps DE Lamarr Houston lines up over the left tackle
According to Pro Football Focus, Houston—Oakland’s top pass-rusher—almost always lines up over the left tackle. That’s good news for Dallas, since the last thing they’d want is Houston on right tackle Doug Free.
When defenses don’t change their alignments all that much, it can become a little easier for offenses to double-team certain players. If I were a member of the Cowboys’ coaching staff, I’d implement more double-tight formations than normal this week, using an extra tight end to help Smith on Houston, combining the run-oriented looks with play-action to attack downfield in non-obvious passing situations.