Unlocking Insights for Fantasy Football Teams with AI-Generated Grades and Explainability
The National Football League (NFL) generates an enormous amount of data, with over 1,700 players participating in 272 games, making it challenging for fantasy football team owners to make informed decisions. The ESPN Fantasy app, which engages 12 million users each year, has partnered with IBM to infuse its fantasy football experience with AI-generated insights that help users make more informed decisions.
Personalized Player Grades and Why
For the past eight years, IBM has collaborated with ESPN to develop a system that provides personalized player grades to help fantasy owners of all skill levels make better decisions. These grades are based on a variety of factors, including the player’s raw performance, their strengths and weaknesses, and how they will complement a user’s fantasy roster.
Making the Grades
The grading system uses a rules-based system in combination with machine learning models to generate grades. The process begins by calculating raw grades based on various factors, including the number of leagues that own a player, the percentage of leagues that start the player, and the player’s projected seasonal stats.
Personalized Analysis at Scale
To provide personalized analysis, IBM’s team needs application personalizes these grades specific to a user’s team every 10 minutes. The application considers the user’s roster, league, and specific settings to deliver customized grades. This process requires significant scalability, as the app receives thousands of hits per second, which are scaled out across pods on a Red Hat OpenShift cluster.
Personalized Explainability at Scale
To provide users with a clear understanding of the reasoning behind the AI-generated grades, IBM has introduced a new feature that unpacks the reasoning behind the grades. When a user taps on a player to acquire or trade, a list of "Top Contributing Factors" appears alongside the numerical grade, providing a personalized explainability in natural language generated by the IBM Granite large language model (LLM).
Conclusion
The combination of AI-generated grades and explainability provides fantasy football team owners with a more informed decision-making process. As in real-world organizations, managers of fantasy football teams need clarity about the "why" behind AI-generated output. The Top Contributing Factors provide explanations based not just on a player’s raw performance but also on the specific ways in which they will complement a user’s fantasy roster.
FAQs
Q: What is the purpose of the grading system?
A: The grading system is designed to help fantasy football team owners make more informed decisions by providing personalized grades based on a variety of factors, including a player’s raw performance, strengths and weaknesses, and how they will complement a user’s fantasy roster.
Q: How does the grading system work?
A: The grading system uses a rules-based system in combination with machine learning models to generate grades. The process begins by calculating raw grades based on various factors, including the number of leagues that own a player, the percentage of leagues that start the player, and the player’s projected seasonal stats.
Q: What is the purpose of the Top Contributing Factors feature?
A: The Top Contributing Factors feature provides explanations based on a player’s raw performance and how they will complement a user’s fantasy roster, giving users a clear understanding of the reasoning behind the AI-generated grades.
Q: How does the team needs application work?
A: The team needs application personalizes grades specific to a user’s team every 10 minutes, considering the user’s roster, league, and specific settings to deliver customized grades.
Q: What is the IBM watsonx AI platform?
A: The IBM watsonx AI platform is a comprehensive platform that helps organizations collect, store, and analyze data, providing timely, up-to-date insights with fine-grained contextual detail.
