False Starts and Snap Counts
FALSE STARTS and COMPLEXITY
After the Seattle Seahawks committed three false starts against the Houston Texans in week four, Pete Carroll promised that his team would work on eliminating this annoying trend. True to his word, the Seahawks committed zero false starts the following week against the Indianapolis Colts.
They also lost.
Committing a penalty is always bad on the play that it happens. But at the same time, an excessive effort to reduce penalties can reduce the quality of play. The theory is logical and straightforward: If your offensive linemen are getting a good jump very close to the snap, then they'll perform better on average while being called for (hopefully just a few) false starts.
More importantly, perhaps, if the snap counts are complex and varied, offensive players will make more pre-snap mistakes but the defense will have a more difficult time anticipating the snap count. So, again, the performance on non-penalty plays improves.
So goes the theory. And going by the eponymous "eye test", the Seahawks seemed to simplify the snap counts when playing the Colts and so allowed the defenders to get a better jump.
I decided to put this theory to a rudimentary test by looking at the sack rates and rushing averages over the course of the season.
First, we assume that false starts represent a more varied snap count. Next, we need to account for other significant factors in run blocking and pass blocking that will cause large swings int he data. So, just to give a basis for comparison, I created an "expectation index" whereby two false starts was equal to the difference between an away game and a home game. I further estimated that the presence of Pro Bowl left tackle Russell Okung was worth 70% of the difference between home and away. Yes, it's completely arbitrary-- but this is just a test.
The "index" then looks something like this:
Home game, Okung playing, 1 false start: 1 + 0.7 + 0.5 = 2.2 (vs. the Vikings)
Away game, Okung out, 3 false starts: 0 + 0 + 1.5 = 1.5 (vs. the Cardinals)
If the theory is correct, then a higher index should correlate with better yards per carry and fewer sacks. Here's a chart of all games on the season, ordered from best-expected performance to worst-expected performance. Opponent quality is accounted for by measuring yards/rush as compared to what each opponent normally allows, and "expected sacks" based on the opponent's seasonal sack percentage and the number of drop-backs:
The offensive line's worst performance of the season-- 7 sacks allowed and just 2.93 yards/carry-- came against the St. Louis Rams, on the road, with no Okung, and zero false starts. But any chance of validating the theory ends there. The contests against the Colts and Texans were both played on the road with Okung out, the only predicted differential therefore being the number of false starts (3 against the Texans and none against the Colts). But the offensive line performance against the Colts' defense was better-than-expected in terms of both sacks allowed and yards per rush. Against the Texans, there was a notable improvement in yards/carry (versus a typical Texans opponent) but that was more than offset by a dismal performance in pass blocking, where the Seahawks gave up 5 sacks (compared to an average Texans opponent, who would be expected to give up just 2 based on the number of times Wilson dropped back to pass).
Sometimes, the numbers show what you expecteds; sometimes they show the opposite; and sometimes, they just look at you and say, "What the hell, dude?"
In that case, a good scientist acknowledges that his hypothesis is a load of hooey and throws it on the back burner.
PLAY CLOCK and ANTICIPATION
A different, but closely related issue in offensive line performance is the timing of the snap relative to the play clock. According to Football Outsiders' Pace Stats, Seattle is the league's 3rd-slowest team on offense. Against the 49ers last Sunday, it seemed that the offense was constantly running late in getting off the plays.
To get a quick-and-easy statisitcal look at possible effects, I sorted the Play-by-play stats from that game and calculated the time elapsed before each Seattle offensive play. This gave no information at the beginning of a drive or following an incomplete pass (because the clock is stopped), but still left 37 offensive plays to examine.
(This includes two plays that came after penalties had stopped the clock; in those cases, I added 15 seconds to the elapsed time, and manually verified on the game tape that this corresponded well with non-stoppage plays.)
The theory here is even simpler: If Seattle is running the clock down, then Wilson has less opportunity to do a hard count and freeze the opposition; when the play clock is close to zero, the 49er defenders can more easily anticipate the snap and get a good jump.
The measurable plays broke down as follows:
41-47 seconds elapsed:
4 rushes for 2 yards (includes a 15-yard face mask penalty; non-penalty plays are 3 rushes for 17 yards)
5/9 passing with 1 sack, 20 yards net (includes a 10-yard offensive pass interference)
37-40 seconds elapsed:
6 rushes for 6 yards (includes a 10-yard holding penalty; non-penalty plays are 5 rushes for 16 yards)
5/8 passing with 1 sack, 47 yards net
27-36 seconds elapsed:
7 rushes for 21 yards
2/3 passing, 13 yards net
A scatter plot, comparing yards gained (y-axis) and seconds elapsed between plays (x-axis) showed a slight correlation:
On average, every 3.5 seconds additional time between plays cost Seattle one yard. The hypothesis is, at least, not proven wrong, although the sample size is far from adequate to prove it correct.
But in that case, we can default to the "eye test" for provisional judgment. And the eye test says that the San Francisco defenders were very aggressive in anticipating the snap, got at least one drive-killing sack when the ball was snapped with under one second remaining, and shut down Marshawn Lynch all day long.
Given that Seattle's offense is one of the league's best on the year, it's fair to say that they perform well given adequate time to call the play and line up. But lining up just a few seconds sooner, thus making the snap less predictable, would probably serve them well against an attacking defensive front.