Fantasy Basketball Assistant
LeGM-CP is a Model Context Protocol (MCP) with a dedicated UI designed to assist in drafting for fantasy basketball. The project started from my own struggles in fantasy leagues, where I found myself consistently losing and realized I needed AI-powered assistance to improve my drafting strategy. By combining projected player rankings, real-time player statistics, and customized punt strategies, LeGM-CP provides data-driven recommendations to help construct a well-balanced fantasy basketball team.
Implementation
LeGM-CP runs on a client-server architecture, with Next.js powering the frontend and the MCP framework handling communication. Page routing is used to connect the two components to allow user interaction and data retrieval.
It combines multiple resources and tools to guide draft decisions, including PDFs of last year’s and this year’s projected player rankings for contextual knowledge, as well as tools that retrieves live statistics player statistics.
A structured prompt template embeds league rules and uses meta meta prompting to generate draft strategies, which is fed through Claude 3.7 Sonnet. This LLM synthesizes stats, rankings, and strategies into actionable draft picks that best fit the user's team.
Conclusion
Although the NBA season has not yet begun and the real-world effectiveness of the drafted teams is still to be determined, LeGM-CP already demonstrates its ability to build well-structured teams that follow logical punt strategies. The project highlights the potential of combining LLM-driven reasoning with structured sports data to improve fantasy basketball outcomes, while also making the drafting process more competitive and enjoyable.
