LEBRON (Line Evaluation Betting Research and Odds Navigator) is a chatbot that uses a machine learning model to predict the outcome of NBA games and give the user advice on how to make favorable sports bets, narrated by none other than LeBron James himself.

It was created in 24 hours for the HackAI 2023 hackathon with Suvel Muttreja and Wei-Chen Huang. I created the chatbot interface and the LangChain agent and integrated the voice narration, Suvel created the voice model and worked on the interface, and Wei-Chen trained and tested the betting model.

View the project on Devpost or the source code on GitHub.


tech stack

I built the chatbot frontend using React and the react-chatbot-kit package and connected it to a LangChain conversational agent which called OpenAI’s GPT-3 API. Suvel cloned LeBron James’ voice using ElevenLabs, and I used the resulting API to add narration to the chatbot interface.

Wei-Chen trained the sports betting model using TensorFlow and attempted to create an interface for it using FastAPI but ultimately didn’t have enough time. Because the chatbot is powered by LangChain, it would be relatively straightforward to add the model as a tool for the conversational agent.