Yes, this is a reference to the dumb 2 pt attempts Buffalo tried, but they are far from the only team. First off, the coaches don't really know how to apply them properly. If they did, they'd understand the data behind them is very incomplete. There's also no really good way to fix it either as there are simply too many variables. Every competent statistician knows when you add or remove variables, you are supposed to start new data sets. All these charts are based on the proverbial 30,000 foot view of the problem. There's no way to capture if a specific guard or center can beat his man, or a receiver runs a crisper pattern, or is more physical than the DB, or is matched up against a LB, whether a LB is good at covering a TE, does a play call provide space for receivers, does a coach do a better job of scouting the opposing team's tendencies, are players getting fatigued, when does fatigue typically set in for certain players, etc, etc, etc. Now, not all variables always matter, but when you have 22 players on a field, with liberal substitution and coaching staff tendencies, it should become apparent the number of variables grows dramatically, and there is simply no way to generate enough statistical data to be truly meaningful.
My solution - make every coach take a course that highlights the limitations and misuses of statistics, and then gather all their stupid charts and throw a White Sox style party at midfield.