We're not talking about whether analytics is the right way to go, we're talking about how to properly apply it. In football, the goal is not to maximize your expected points: the goal is to have more points than your opponent when the clock runs out. The two goals are not the same. You must apply the math/analytics in the correct context. You must understand that information has value. You used poker as an example: information is key to the strategies in that game,and the correct strategies change as each card comes down and you acquire more information (thinking about Texas Hold'em and similar games in this case). In football, the clock and the current score matter: your decisions are not independent of those factors.Holy cow. If you're down 6 and score a TD as time expires, the game is tied and you have one untimed extra-point play left. If you kick the extra point, there's a 97% chance you win and a 3% chance you lose (assuming OT is 50-50). If you go for 2, there's a 74.25% chance you win and a 25.75% chance you lose. If you can't figure out which the better play is there, I don't know how we can have a serious discussion about it.
Similarly, if you are down 6 and score a TD anytime in the last let's say 5 minutes, the 1-pointer is clearly the better play. Why? Because being ahead by 1 point or 2 points are identical if the opponent can get only one more possession: they beat you with a FG. The second point adds virtually nothing to your win expectancy, but missing the 2-pointer and getting 0 points has a massive effect on it. You have sufficient information to make a more informed choice than just "the 2-pointer is the higher EV play." And yes, you can work this out mathematically, but you must factor in information such as the likelihood of the other team scoring with the time left, the likely length of such a drive, etc. The EV of that particular play is only one piece of data.No, that's not at all why analytics exist. It exists to find the best strategies and tactics that maximize your chance of winning the game, not scoring the most possible points over 3 seasons worth of games. Let me give you an example. I'm ahead by 2 points, there are 100 seconds left in the game, I have the ball 1st-and-goal at my opponent's one-yard line, and my opponent has no time outs left. I basically have two choices. Take a knee for 3 straight plays, or try to score the TD. Trying for the TD clearly has a higher expected points value for me than taking a knee. It is also clearly the wrong play, because it increases the chances of losing the game (via turnover). The tail of the distribution where I lose is much much bigger than if I take a knee. It's the distribution of outcomes that matters, not the expected value of the play.
Actually, there's a third option in the scenario I just described: kick a FG on 1st down. That play has a much higher EV than taking a knee, but it's just about the worst possible thing I could choose to do. I just gave the opponent back the ball, down 5 points, with 90+seconds left. It's an awful idea despite the EV, because I failed to take the clock and the current score into account.
