Is it possible that analytics influenced SC decision on that play?
I don't think it's terribly likely that they would have actually favored letting Alabama score a touchdown.
I don't know the exact odds, but a field goal attempt from that range should have been around 95%. That's a 1/20 that they miss. This without factoring in the odds of a bad snap/fumble/penalty. So the downside is Alabama almost certainly runs out the clock if you don't stop them in the field of play. The upside is you probably have about a 5-10% chance of stopping a score if you get him down in the field of play given all the variables.
On the other hand, 34 seconds, no time-outs, needing a touchdown to tie the game. I think that's a very low probability. For the record I decided to run this exact scenario through AI, which isn't perfect but is also similar to the way analytics would work. They came up with a 2% chance of success if South Carolina lets Alabama score, and when it used my estimate it was a 7.5% chance of success (5-10) if they tackled the Alabama player. When I made it come up with the own numbers, it was 8%.
So basically, you are four times more likely to succeed in that scenario if you don't let Alabama score. However, when you talk about the difference in 8% chance and 2% chance, I think a lot of humans are not likely to really grasp that difference as being as significant as it is. I don't see how any well calibrated analytic would say let Alabama score though, not with no timeouts.
If they had over a minute and timeouts, their odds of scoring a touchdown go up to about 15%,, then it becomes the right call. Alabama could have drained the time/timeouts with it being first and goal from the 5 and that is exactly the scenario where teams let them in the end zone, So, basically I think the player and/or coach could have misunderstood the analytics in this scenario.
