Jump to content

Matt Schaub made a good decision.

With 2:51 to play and a seven-point lead, Matt Schaub threw a heartbreaking pick-six. The results were bad, but it was worth the risk. Image from The Gazette
If you don't have a lab full of monkeys, you should get one. They provide valuable insight into cognitive bias. One such bias is loss aversion. It is an error in judgment whereby something that is already possessed is overvalued relative to something that might be acquired.

Case in point: a fourth-quarter lead. Football dogma says that a team that is winning by more than a field goal late in the game "should" win. The offense needs only to kill a bit of clock and avoid a turnover. The defense "should" hold.

Make no mistake about the implied responsibility: A defense that gives up a touchdown in this situation is considered an utter failure, on par with an offense that lines up in an illegal formation, an untouched ball carrier fumbling, or a player getting ejected for throwing a punch. Professionals simply don't do that sort of thing. It does not require excellence nor even superior performance, just basic competency.

We expect a defense in that situation to gain some mystical advantage by having the final victory in sight and needing just one more effort. In fact, the opposite is true. A defensive stop for the win on the final drive of regulation is much harder than any other defensive stop, because your opponent is willing to take risks and has all four downs to work with.

Last Sunday, Houston took possession at their own 43 with 5:13 to play and a 20-13 lead. At this point, they have five paths to victory:
  • Give the ball back and prevent an opposing score
  • Run out the clock
  • Score on this drive and take a 10+ point lead
  • Give the ball back, allow a touchdown, and then score again
  • Give the ball back, allow a touchdown, and win in overtime
Any suggestion to an average football fan that those last two items be given serious consideration in formulating a strategy is likely to meet with exasperation. It's like suggesting that one possible way to travel to the supermarket is to drive your car part of the way, destroy it in a wreck, and then take a cab the rest of the distance.

The intermediate setback so wrong, so disastrous, so incomprehensible that no thought should be given to such contingencies. It should be avoided, must be avoided, at all cost.

That's loss aversion at work.

But now comprehend the reality:

Consider the 4th quarter action in each of the last three games where a team managed to beat the Seattle Seahawks. The Dolphins gave up a 4th quarter touchdown and had to score 17 4th-quarter points themselves. Detroit blew a 4-point lead before scoring the winning touchdown with 20 seconds to play. Atlanta blew a 20-point fourth quarter lead and needed a field goal with 31 seconds left in the game.

Other Seahawk opponents have not fared so well in the Russell Wilson era. The Packers, Patriots, Bears, Redskins, and Panthers all had fourth quarter leads. Each of these teams punted the ball, in the fourth quarter, with a lead. Each of these teams lost.

Was the Seattle-Houston matchup so different defensively, with the Seahawks being held to just 13 at that time? Well, Seattle had only 13 points through three quarters against the Redskins, and no more than 10 points after three quarters in any of the other games.

Matt Schaub's moment of destiny came with 2:51 to play, facing 3rd and 4 on the Seattle 40. The clock is ideal for a team to move the length of the field and score, providing plenty of time to run a no-huddle offense but brief enough that it would likely be the final drive in regulation. Historical data says that a failed 4th down conversion here will lead to a touchdown by the opposing team about 45% of the time. Russell Wilson says he can beat that percentage.

If a pick-six ties the game (as happened), historical data says that the Texans have about a 33% chance of closing it out in regulation. Despite needing only a field goal in that situation, the percentage is lower than that of the TD drive for obvious reasons: A tied team will not take risks and use all four downs as would a team trailing by a touchdown.

If the Texans convert for a first down, it's almost a guaranteed win (they'd be in long field goal range). Overtime in any case is a 50-50 proposition.

With all that in mind, the historical projections of winning percentage (WP) can be found by plugging the game state into Advanced NFL Stats' Win Probability Calculator. For your enjoyment here are the play results and Win Probabilities for the Texans drive prior to the interception:

* play: start of drive (result: Q4 5:13, 1st and 10, HOU 43) … WP = 0.94
* play: Foster 6 yard run (result: Q4 4:27, 2nd and 4, HOU 49) … WP = 0.92
* play: Foster 5 yard run (result: Q4 3:43, 1st and 10, SEA 46) … WP = 0.96
* play: Foster 5 yard run (result: Q4 2:58, 2nd and 5, SEA 41) … WP = 0.93
* play: Foster 1 yard run (result: Q4 2:51, 3rd and 4, SEA 40) … WP = 0.87

Then, at the critical juncture, here are Houston's win probabilities for each of several possible outcomes:

* play: Incompletion/ no gain (result: Q4 2:47, 4th and 4 , SEA 40) … WP = 0.77
* play: First down (result: Q4 2:47, 1st and 10, SEA 36) … WP = 0.97
* play: Interception with no return (result: Q4 2:40, 1st and 10 Seahawks, SEA 40) … WP = 0.83
* play: Pick Six (result: Q4 2:40, Seahawk TD, Houston's ball after kickoff) … WP = 0.67

Notice that the difference between converting a first down and giving the Seahawks the ball back is HUGE. Houston wins 97% of the time if they convert, and just 77% of the time if they do not.

Now observe that Houston's win probability after a pick-six is really not that bad-- still 67%. They can win it in regulation with just a field goal drive or they can win in overtime.

So now let's analyze Schaub's decision to go for the first down. For the sake of simplicity, let's assume that he has two choices: Throw the ball away (guaranteed incompletion) or throw it to Owen Daniels.

Furthermore, let's assume that a pass to Daniels has just two possible outcomes: Either Daniels gets the ball and the first, or Richard Sherman picks it off and scores a touchdown.

How much of a chance does Schaub need to make this a smart pass?

Some accused Schaub of throwing up a "50/50" ball, but, surprisingly, his chances didn't even need to be that good. The "break even" point compared to an incompletion is a paltry 33%, as demonstrated:

33% success rate X 97% WP if successful = 32% partial WP
67% failure rate X 67% WP if intercepted for touchdown = 45% partial WP
Total WP for risky pass = 32% + 45% = 77%

Watching the replay, Schaub had two receivers to the right (Andre Johnson and Owen Daniels) covered by Kam Chancellor and Richard Sherman. Chancellor blitzed. Basic math puts one DB on two receivers. Fundamental quarterbacking rules say that you throw to the blitzing side. Common sense says the lone defender ought to be covering the deep man, especially if his name is Andre Johnson.

The replay shows that Owen Daniels had position, making this less "throwing up a prayer" and more like a low-post lob to your power forward. Daniels not only got out-muscled for the ball, he failed to react in time to Sherman's presence, missing an opportunity to break up the pass or, at the very least, make a tackle.

The numbers show that Houston had good reason to take some risk. It's a smart pass even when the probability of a pick-six is much higher than the probability of a first down.

As a Seahawk fan interacting with Texans followers on various forums; and watching the home crowd response to the Michael Bennett injury; I have to say that this is a very classy fan base.

Watching Matt Schaub shred the league's best pass defense in the first half, I have to say that the man has some mad skillz. And he doesn't exploit them by taking sacks and throwing the ball into the turf.

So I implore you not to judge Schaub so harshly for what, as the numbers show, was a smart risk. Yes, he gets paid to take the verbal abuse and effigial burnings, but tearing your hair out over a results-based analysis isn't going to make anyone happier.

Share on Fancred


Seth Keysor
Oct 05 2013 06:49 PM

Really nice piece.  It's easy to get caught up in hindsight and results, but to an extent the throw makes sense.  They got burned by the worst possible outcome.

Jason Drake
Oct 05 2013 08:18 PM

... They got burned by the worst possible outcome.

True. But, and I can't stress this enough, that wasn't the winning play. Houston still had the better chance to win.

Curiously, both offenses were very reliable at moving the ball (if not necessarily scoring) from that point on:

Q4 after pick:
Houston 7 plays, 37 yards (punt from opp 43)
Seattle knee

Houston 5 plays, 20 yards (punt from own 40)
Seattle 4 plays, 33 yards (punt from opp 47)
Houston 7 plays, 37 yards (punt from own 40)
Seattle 8 plays, 42 yards (FG from opp 27)

The second and third drives of OT (Houston for 33 & Seattle for 37) were both good enough to get into field goal range from better field position, but the previous opponent's drive gave them too much field. So you could argue that it was effective offense, as much as effective defense, that dragged out the overtime period to 11 1/2 minutes.

    • Seth Keysor likes this

Fantasy Football Articles