It only works if the analytics are interpreted and applied correctly.
In the article you reference, they only used data from the 1st and 3rd qtrs, eliminating half the game and arguably the quarters that going for it would be more likely to be used. They also eliminated data from a wide range of factors. Stats work because it measures the probability of an outcome over time, but the frequency of that result is not going to fall into that same probability. Each attempt is independent of each other, just because you miss 3 attempts in a row doesn't mean the next attempt will be successful. I have been in a casino and watching roulette where black hit for 14+ times in a row on numerous occasions, eventually it would regress to the mean, but it also doesn't mean it would regress the same way. Instead of 14 red in a row, it may be 3 or 4 in a row over a period of time.
Those stats are also very general and would have to be recalculated for each team and individual factors like opponents, weather, surface, personnel, time remaining etc. While the success is measured over time, each game has a finite amount of time. Analytics are useful, but again, it's about interpreting and applying the information. It requires a feel for the game and moment and calculating the risk involved. If you fail on successive attempts, it will change the risks you may have to take to overcome them. You have failed to score points, turn over the ball giving the opponent opportunity to score and used time.
