I'm not sure I am buying the scientificity of this simulation. If it draws its probabilities from plays in actual NFL games since 2008, then it probably doesn't just look at plays in the last four minutes in close games. In other words, it probably presumes that any given play is equally likely to be successful (or not) in the first quarter as in the fourth quarter. It does not and could not take into account the possibility that what is a high percentage completion in the first quarter might be less high percentage (because more contested) in the closing minutes of a tight game. And if completion percentages are down then the argument in favor of four minute offense grows stronger in comparison.