Filed under:

# Harsh judgment for Houston scapegoats

Matt Schaub and Kareem Jackson are victims of bad timing and perception.

What do you call a batter who hits into a double play, strikes out, then caps off the night with a pair of home runs to win the game?

Clutch

What do you call a batter who starts with a pair of home runs, later strikes out, and finally hits into a double play in the bottom of the ninth with his team down by one run?

Choker

I use a baseball metaphor here because fans are more familiar with its large sample sizes that produce every possible result. With 162 games and 600+ at-bats, a hitter's performance in high-leverage situations will eventually reflect the same averages as his overall performance.

Not so in football, where a single play can determine the outcome of a game and a single game can determine the outcome of a season. But even the best players make bad plays. Success requires risk. And with the complex nature of play calling, there is a significant random component; while better teams and better players will prevail over many iterations, any one play can go to hell if the opposing stunt/coverage/screen/draw comes at the wrong time.

For that matter, consider free throws in basketball. Shooting 85% is considered excellent, but still encompasses a 15% failure rate, despite every attempt being absolutely identical and unaffected by opposing defense. If a player could really will himself to an extra effort and make "clutch" free throws, why would he miss any?

### Matt Schaub's Pick-Six: Disaster or Setback?

I used Advanced NFL Stats' win probability calculator to take a closer look at the Texans' second-to-last drive in regulation last Sunday. With Houston up 20-13, here are the plays that preceded Richard Sherman's interception, and the corresponding probability of a Texans win after each play:

Play Time Result WP
Q4 5:13 1st and 10 HOU 43 0.94
Foster 6 yard run Q4 4:27 2nd and 4 HOU 49 0.92
Foster 5 yard run Q4 3:43 1st and 10 SEA 46 0.96
Foster 5 yard run Q4 2:58 2nd and 5 SEA 41 0.93
Foster 1 yard run Q4 2:51 3rd and 4 SEA 40 0.87

Note that Houston's win probability shrinks on later downs despite advancing the ball. That's because of the game clock. When there were five minutes to play, Houston had four paths to victory in regulation:

#1 score on this drive (go up by two scores)
#2 run out the clock
#3 give the ball back and hold on defense
#4 give the ball back, allow a tying score, and then win on a final drive.

The minimal gains don't much help Houston's chance of scoring, and by the time they have 3rd and 4 at the Seattle 40-yard-line the game clock is at at ideal one-possession time. Really, ideal. I plugged several scenarios into the WP calculator and played with the times to confirm this. Two minutes and change is the perfect amount of time for a final drive: a team can use the whole field, call audibles, and (if necessary) manage the clock in the red zone to ensure that their opponent has no final opportunity.

At this critical 3rd and 4, Houston has already lost one scenario for victory in regulation (an exchange of scores). If they fail to get a first down, they will have lost two more (running out the clock or scoring). Arian Foster has carried the ball for four straight plays and the Seattle defense is bearing down to stop the run, making a rushing first down highly improbable. There is a lot of value to attempting a pass on this next play.

Some of the possible results, and Houston's corresponding win percentage, look like this:

Play Time Result WP
Incompletion/ no gain Q4 2:47 4th and 4 SEA 40 0.77
First down Q4 2:47 1st and 10 SEA 36 0.97
Interception/ no return Q4 2:40 1st and 10 Seahawks SEA 40 0.83
Pick Six Q4 2:40 Seahawk TD n/a 0.67

A funny quirk here is that an interception appears to actually be better than an incompletion. You could argue that the WP model is broken, but I don't think so. Remember that it's based on historical data and so affected by coaching decisions which may not be ideal. There are undoubtedly some blocked punts, big punt returns, and a few ill-advised 57-yard field goal attempts.

More importantly, a punt here is virtually worthless. At two minutes and change, a team with 80 yards to go for a tying TD is no worse off than a team with 60 yards to go for a tying TD. Those are just about the easiest 20 yards of field position to get in football, and they come with the added bonus of getting into an offensive rhythm and wearing down the opposing defenders. Again, this is not a subjective opinion. It's based on historical data that informs the WPA model.

Back to the play, then:

Matt Schaub drops back, turns, and disovers Kam Chancellor looming. Let's assume his choices are to throw a deliberate incompletion or make a risky throw; furthermore, just for kicks and simplicity, let's assume that the risky throw has only two possible outcomes: a first down or an interception for a touchdown.

It's then a simple matter to calculate what the success/failure rate on the risky throw needs to be in order for it to be as good a decision as throwing away the ball:

Throw it Away Risky Pass
WP Success: .77 .97
Success Rate: 1.00 .33
Net WP success: .77 .32
WP Failure: n/a .67
Failure Rate: 0 .67
Net WP failure: 0 .45
Total WP: .77 .77

Your eyes do not decieve you. The calculated break-even point to make the throw is one chance in three of a first down, and a two-thirds chance of a pick-six. Remember the clock situation! Giving Seattle the ball back puts them in the ideal time frame for a final drive to tie the game; and giving up an interception TD return puts Houston in that same time frame needing only a field goal to win (and still having a 50% chance of winning if they merely run out the clock).

So, to recap Matt Schaub's day:

• 355 yards passing, 5 yards rushing, 2 touchdowns, 2 interceptions
• Individual WPA = +0.16, 10th highest among quarterbacks for the week against the league's best pass defense
• WPA from running plays on critical drive: -.07
• WPA from interception on that same drive: [no worse than] 0.00
• WPA from tight end Owen Daniels failing to tackle Richard Sherman after getting mugged: -.20
• ### Overtime Body Slam

Finally, let's take a look at the WP impact of Kareem Jackson's personal foul penalty on Seattle's final overtime drive (WP is for the Seahawks):

Play Time Result WP
OT 6:46 1st and 10 SEA 31 0.57
Lynch 6 yard run OT 6:16 2nd and 4 SEA 37 0.62
Wilson 7 yard run OT 5:42 1st and 10 SEA 44 0.67
Lynch -2 yard run OT 5:00 2nd and 12 SEA 42 0.63
Baldwin 7 yard rec OT 4:52 3rd and 5 SEA 49 0.65
15-yard penalty OT 4:52 1st and 10 HOU 36 0.78
Lynch 3 yard run OT 4:16 2nd and 7 HOU 33 0.81
Lynch 6 yard run OT 3:28 3rd and 1 HOU 27 0.85
Incomplete pass OT 3:23 4th and 1 HOU 27 0.81
45 yard field goal OT 3:18 Seahawk FG n/a 1.00

Going from 57% to 100%, Seattle had a +0.43 WP on their final drive. Jackson's foul was worth +0.13 to the Seahawks, only a fraction of the total and not even the biggest single play (Hauschka's field goal was worth +0.19).

I'll admit this bugs me as a Seahawk fan more than as an apologist for Kareem Jackson. The highlight programs all showed the penalty, ignoring the tough running that made up the bulk of the drive. Lynch's last run of six yards came with ten men in the box, five of whom were required to bring him down.

Seattle really needs to lobby for a rule change making it an Unnecessary Roughness penalty to jump on a ball carrier when there are already two guys hanging off him.

That would be totally unfair, of course.

Which is the point, of course.