Considerations for Mark Nobel Contract - Oakland A’s Ticket Sales Model

Considerations for Mark Nobel Contract

Oakland A’s Ticket Sales Model

In order to best approximate the value of Mark Nobel, a model must first be established for baseline Oakland A’s ticket sales. Building the model requires the following adjustments to available data.

1. Variable Adjustment

Of the available game variables, the date of the game can be logically removed – all aspects of the dates that are evaluated on a continuum (weather, importance of game, etc.) are adequately represented through other variables.

There reaches a point in the season where the A’s no longer have a chance at making the playoffs. A new variable has been created (“Elimination”) for when the approximate number of games remaining (home games and double headers left multiplied by two) is greater than the number of games behind 1st place the A’s are at a given point.

Day-of-week-adjusted-attendance (AttAdj) has been created based on the following Daily Index:

1. Variable type definition

The following variables are continuous: AttAdj, Position, Gamesback (games behind 1st), and Temperature. The following variables are nominal and thus must be represented as a separate variable for each possibility: DH (doubleheader), Opponent, Elimination, Precipitation, Time, Televised, Promotions, and Nobel.

An initial regression in JMP 13 (Exhibit A) suggests that the following variables are insignificant:

• Televised – This variable is logical to remove, as the decision to attend a baseball game is made by many factors beyond whether a game is available on television.
• Time – It appears that factors in favor of night games over day games, or day games over night games, are insignificant
• Nobel – Most notably, the Nobel variable did not appear significant. Because this analysis centers around Nobel, it will be left in the model.
• Position – Perhaps given that opponents are roughly synonymous with positions in the standings over the course of a season, the opponent’s position in the standings was no significant over one season of data.

Instead of using a more typical significance threshold of PValue <= 0.05, a value of <0.30 will be used so as to not eliminate variables that have a reasonable chance of being significant in different baseball seasons.

The final model is shown in Exhibit B. Below are some significant model effects. Note that extra fans for certain opponents are measured in comparison to the team that has the least positive effect on attendance (Texas).

Impact of Mark Nobel

According to our model, on games that Mark Nobel was pitching, attendance was actually 325 less fans than was to be expected at those games. The fact that the Nobel variable was not significant in our model indicates that it’s actually more likely that Nobel had no effect in attendance. To directly retort Nobel’s agent’s claim of attendance figures, the A’s were actually supposed to have 12,988 attendees in games that Nobel started.