OK, first of all, this whole discussion hinges on the assumption that in the remainder of the game, we manage to outscore the other team by exactly one TD. If we don’t do that, it doesn’t matter what we chose, because we’ll lose regardless. If we get more than one TD (a TD and a field goal, say), then we win regardless. There is some probability, P[sub]T[/sub], that we’ll get exactly one more TD, and another probability, P[sub]F[/sub], that we’ll get more than 1. We don’t know what these probabilities are, but they won’t change which choice is better. There’s also a probability of success on a point after, P[sub]1[/sub], which I’ll call .95, and a probability of success on a 2-point conversion, P[sub]2[/sub]. In addition, there’s the probability that we’ll win if we go to overtime, P[sub]O[/sub], which should be .5 (since, in every scenario where we go into overtime, there’s no way to distinguish between the two teams).
All right, then. The two strategies to consider are strategy A, where we try to make a point-after after each TD, and strategy B, where we first try to get a two-pointer. Under strategy A, we try to make the point-after right now, and then another point-after if we succeed, or a two-pointer if we fail, tying the game either way, and then hope to win in OT. This makes our probability of winning
P[sub]A[/sub] = P[sub]F[/sub] + P[sub]T[/sub]*( P[sub]1[/sub]^2 + (1-P[sub]1[/sub])*P[sub]2[/sub] )*P[sub]O[/sub]
In strategy 2, we first try for the 2, and if that succeeds, we try for the 1, and if it fails, we try for 2 again and hope to win in overtime. That makes our probability of winning
P[sub]B[/sub] = P[sub]F[/sub] + P[sub]T[/sub]*( P[sub]2[/sub]*P[sub]1[/sub] + (1-P[sub]2[/sub])*P[sub]2[/sub]*P[sub]O[/sub] )
Comparing these two, we see that P[sub]F[/sub] and P[sub]T[/sub] don’t matter. We have to compare
( P[sub]1[/sub]^2 + (1-P[sub]1[/sub])*P[sub]2[/sub] )*P[sub]O[/sub] >?< ( P[sub]2[/sub]*P[sub]1[/sub] + (1-P[sub]2[/sub])*P[sub]2[/sub]*P[sub]O[/sub] )
Plugging in our values for P[sub]1[/sub] and P[sub]O[/sub], we have
( 0.45125 + 0.025P[sub]2[/sub] ) >?< ( 0.95P[sub]2[/sub] + 0.5*(1-P[sub]2[/sub])*P[sub]2[/sub] )
I don’t feel like solving the quadratic to find the break-even point for P[sub]2[/sub], but assuming it’s 0.45, then we have
0.4625 >?< 0.59875
which strongly favors strategy B, going for the two-pointers. Assuming it’s 0.4, then we have
0.46125 >?< 0.5
which still favors B.
Quoth whole bean:
Because you don’t have any momentum to kill. The statistics are clear that “scoring momentum” exists only in the minds of coaches and fans: It does not show up on the field. And coaches who make decisions based on things that don’t matter on the field really ought to be fired, though of course they aren’t. The real problem is that the skill set that would make for a good coach is completely different from the skill set that’s actually used to select a coach. Most coaches are former players, but if players were able to make the hard decisions, then we wouldn’t need coaches in the first place.