This is a review of the MLB research conducted by Joseph Xu, Peter Fader, and Senthil Veeraraghavan, applying negative binomial regression.
As much as sports teams are about winning championships, the aim of management and stakeholders is t o maximize revenue. Many teams across various sporting leagues have adopted a dynamic pricing strategy for single-game tickets. However, these prices are typically based on recommendations from outside vendors. The question becomes what pricing strategy will optimize ticket sales and maintain a strong fan base.
In an attempt to answer this question a demand model for single-game ticket sales is developed. Information was generated from one MLB team to test this data against actual ticket sales in order to set appropriate parameters. The team’s pricing strategy is analyzed and three alternative strategies are developed.
The MLB team providing the data instituted two different pricing strategies over the season. Prior to the All Star break, they maintained a variable pricing strategy where ticket prices varied across games and seat sections but remained constant over time. After the All Star break a dynamic pricing strategy was implemented. The data from the first part of the season was used to determine appropriate parameters while the data from the second part of the season was used to create revenue predictions.
Customer demand for single-game tickets is modeled in three stages: game decision, ticket quantity decision, and seat section decision. Game demand is modeled with a negative binomial regression taking into account the effect of time, game characteristics, team performance, price, and occupancy. The resulting parameters illustrated the fact that higher prices result in lower demand, promotions increase demand, and opponents affect demand as well as stadium occupancy and team performance.
The ticket quantity decision, or the second stage of the customer demand, is modeled with a negative binomial regression using the same set of variables as in the first stage. The resulting parameters are the same as those generated in the previous stage.
The third stage, or seat section choice, is modeled using multinomial logistic regression. This model includes additional factors such as the time until the game and the number of tickets required. The resulting parameters included the fact that the best seat sections tend to sell sooner than other seat sections and customers looking for more tickets typically do not purchase the most expensive seats.
These three models are combined in order to make predictions for expected daily revenue for each seat section for every game. Three alternative dynamic pricing strategies are tested using the model, optimal variable pricing looking ahead at team performance, monotone myopic dynamic pricing, and unrestricted myoptic dynamic pricing. The greatest positive change in revenue was created with the unrestricted myoptic dynamic pricing.
Teams can use this information when setting single-day ticket prices that will maximize revenue. Ticket prices should not be changed early in the season but rather closer to the end of the season when demand tends to rise. Not all seat sections need to follow the same ticket pricing strategy. The optimal strategy for each section varies depending on factors like time until the game, team performance, and number of seats remaining in the section.
However, it is important for teams to always keep in mind their fan base – maximizing ticket revenue should not come at the cost of angering and losing the fan base.
Analytics Used: Demand Model, Negative Binomial Regression, Multinomial Logistic Regression
Find out how Sports Analytics Expert Victor Holman can give your team the competitive advantage.
How mature is your team’s analytics program? Take the Sports Analytics Maturity Assessment.
Discover the Groundbreaking Sports Analytics Application and Framework coaches and sports analysts are talking about!
Learn all about sports analytics in Victor Holman’s Sports Analytics Blog.