Methods of using multiple regression in football tendency analysis

A method of analyzing the play calling tendencies of an American football opponent, comprising:

• defining at least three distinct, non-overlapping categories of plays that the opponent might call, all such categories together being exhaustive of the universe of plays that the opponent might call;
• collecting information on the plays that the opponent actually called in prior games, assigning each such play to exactly one of the defined categories, and collecting information about the score, time remaining, down and distance, field position, personnel package, and other relevant information in the prior games when each such play was called;
• inputting all such information collected into a computerized statistical database;
• for each of the at least three defined categories of plays, performing a multiple regression wherein each observation is a play that the opponent ran in a past game, the dependent variable is a binary variable indicating whether the play that the opponent ran was in that category, and the explanatory variables reflect the game conditions during that play;
• using appropriate explanatory variables reflecting the game conditions, to estimate the probability that the opponent will run a play in that category given various game conditions;
• programming a computer to use the coefficients estimated by each such regression to model the probability that the opponent will run a play in the corresponding category under particular conditions during a game;
• entering sets of game conditions into the computer to use as or to calculate values for the explanatory variables of the models;
• and
• solving the models for each category of plays and displaying each category, along with the predicted probability that the opponent will run a play in that category, sorted by probability, from highest to lowest.