Last night, the New England Patriots snatched victory from the jaws of defeat, intercepting a pass on their own goal line with just 30 seconds to spare. Before the play, Advanced NFL Stats gave the Seahawks an 88 percent chance of winning, as most teams end up converting 2nd-and-1 into a touchdown in that situation. Of course, most teams run the ball into the endzone from a yard out, and in the aftermath of the game, the primary focus was on Seattle’s decision to throw the ball rather than run it.
On both NBC and Twitter, incredulity seemed to be the most popular reaction. The overwhelming consensus appears to be that the Seahawks screwed up, and simply should have run the ball. The tenor of the commentary suggested that any other play call was demonstrably wrong; 2nd-and-1 from that spot on the field, with a running back like Marshawn Lynch, throwing the ball should apparent not have even been a consideration.
Of course, the reaction is inextricably tied to the result. Had the play worked, we’d likely be spending this morning hearing about what a genius Pete Carroll is for his creative play calling, zigging when the Patrios expected him to zag. This isn’t to say that throwing the ball was definitively the right decision, but I’m naturally wary of analysis that suggests that there is ever only one correct strategy to deploy, with no other decision ever being reasonable to consider. But because I don’t know football well enough to have an informed opinion, I started wondering what the baseball equivalent would be to the Super Bowl’s ending.
My initial reaction was to think of something like a dominant closer throwing a rarely used pitch in an 0-2 count in the ninth inning of the final game of the World Series. This is the element of surprise play, attempting to trick an opponent with a play they aren’t expecting, rather than simply attempting to beat them with your best weapon. And we actually just saw the Giants do exactly this two years ago.
“That guy,” Posey said an hour later, shaking his own head in amazement. “He shook to a fastball there. That shows the type of guts he has and faith in what he’s got. It’s just a great job by him. This is not a knock, but he throws 88 or 89, but he’s got a plus, plus slider.”
Everyone in the world, Cabrera included, was expecting a slider. Romo has a great slider and a meh fastball. As Jeff noted in that post, Romo had thrown a slider 85 percent of the time in two strike counts that year, and had gotten even more extreme with his pitch selection in that World Series. He had an unhittable slider, got into a slider count, and then instead chose to throw Cabrera a hittable fastball instead. The element of surprise worked; Cabrera took strike three and the Giants were the World Series champions.
But the more I thought about that situation, the less comfortable I was with the analogy. The Giants were trying to win the World Series, but they had no chance to lose it if things went wrong. They were up three games to none at that point, and holding a 4-3 lead in the 10th inning of Game Four. Even if Cabrera crushes the fastball for a home run, it’s only a tied game, and the Giants would get to keep playing. And even if they lost that game, they’d have had three more chances to win one game. Romo’s fastball didn’t give the Tigers a chance to win the World Series had it gone wrong, and if you change the stakes to Game Seven, maybe he doesn’t throw that pitch.
So maybe we need to find a comparable baseball situation that isn’t a closer protecting a lead. After all, while win probability had the Seahawks as the favorites at that point, they were still losing, and failure to score meant that they didn’t have another chance. Perhaps we need something like the tying run on third and the winning on second base, with nobody out, and a great hitter at the plate. According to the Win Expectancy charts, that situation results in the team down a run winning 71% of the time, so to get up to the 88% mark from the Super Bowl, the batter/pitcher match-up would have to really favor the hitter. To tilt the numbers even farther in favor of the offense, we probably need to put a runner on first base as well — taking away the intentional walk and raising the initial Win Expectancy to 74% — and then also start the count in the hitter’s favor; we’ll say 2-0, since failure on the next pitch would have left two more opportunities, just as it would have for the Seahawks.
So, that’s perhaps a comparable situation from a win probability standpoint. Bottom of the 9th inning in Game Seven of the World Series, down a run, but with the bases loaded and nobody out, and your best hitter at the plate, with the count already worked 2-0 in his favor. But what’s the surprise play there? The defense wouldn’t be expecting a bunt, but a bunt doesn’t present a win-or-lose dynamic, as the runner on second base isn’t scoring on anything short of a defensive debacle. Stealing home would also be a wildly unexpected attempt, but again, that doesn’t offer the win-if-it-works incentive, and the rally doesn’t end if it fails.
Instead, you need some kind of strategy that results in both the runners on second and third scoring on a successful execution, but the inning ending on a failure. Basically, you need the downside to be a triple play.
And that means a hit-and-run. All the runners take off with the pitch, and the guy at the plate is told to simply hit the ball on the ground; if it finds a hole, the runner from second scores easily and they win, and even if it ends up going right at a fielding, sending the runners means that they can’t get the force at home, so they’d get to send another hitter up with a chance to drive in the go-ahead run from third base and less than two outs. That chance would be wiped out if the batter hit a line drive right at an infielder near a base, turning one out into three outs in short order. Maybe it would look something like this.
From a win probability and risk/reward standpoint, this probably comes close, and it really is hard to imagine a manager calling for a hit-and-run in that situation in the World Series.
But when comparing this situation to the last play of the Super Bowl, this does fail on one important element of the Seahawks play call: the match-up variable. The Seahawks specifically chose to throw the ball in response to the Patriots defensive package, and the post-game rationale was essentially a game theory argument. If everyone in the world thinks you have to run there, then throwing the ball has a higher probability of success than it would on any random play during the game, since the defense is specifically aligned to stop a run play.
You don’t get that on a hit-and-run, really, because it’s not designed to exploit the flaws in a defensive alignment. The best baseball equivalent of throwing the ball against a goal line defense is probably something like bunting against the shift, which is not at all a controversial strategy when its deployed. It’s pretty widely accepted that if a team is overstacking their infielders on one side of the ball, a good hitter should attempt to go the other way, even if he’s a dead pull hitter whose strength is hitting the ball hard to the pull field.
From a strategic perspective, a better fit would be bunting against the shift, but there are very few plays in baseball where a bunt can decide the game in favor of both teams at the same time, depending on the outcome. And, of course, teams don’t overshift the right side of the infield when there are runners on second and third.
So perhaps we back to the beginning. Perhaps the closest corollary we could get is if Romo’s fastball to Cabrera came in Game Seven, and if there had been a runner on first base, so that a home run from Cabrera allowed the Tigers to walk off as World Series champs.
If the World Series had been on the line, would it still have been wrong for Romo to go for deception rather than playing to his strength? Was it clearly the wrong call for him to throw Cabrera a middle-middle fastball rather than his dominant slider because of the risk of Cabrera hitting it to the moon? Or was it the right call because the stakes were lower for the Giants in that situation?
I don’t think I have an answer. Instinctually, I’m against the idea that there is ever only one correct decision to make, even if your opponent knows that’s what’s coming. But on the other hand, if you have a dominant weapon that almost always works in spite of the lack of surprise, is the final play of the championship game really a time to be messing with game theory, or should you just let your best go up against their best and accept the result?
I’m probably a bit more sympathetic to the Seahawks play call than the national response, but I don’t know football or game theory well enough to say that it was justifiable. I do wonder, though, how different Romo’s decision to throw Cabrera a fastball really was. The circumstances weren’t exactly the same, but the concept was similar. Do we think Romo’s pitch selection was insane, even though it worked? Or are we just allowing the result of the play to define whether or not the idea was sound?
Dave is the Managing Editor of FanGraphs.