basketball-analytics

Kinetic Adaptation via Wasserstein Heuristics and Identity

A curated marketplace of high-quality community sports plugins for Claude Code, Codex, and AI coding assistants. Covers analytics, live scores, fantasy sports, fitness tracking, and more.

Competition-winning Rebound Prediction Notebook

Basketball player tracking with jersey-number OCR and real-time tactical court projection (RF-DETR, MCByte, SAM, PARSeq)

Coaching Experience's Effect on Winning

Basketball analytics for the Lebanese Basketball Federation Division A: scrapes, cleans, and analyzes RealGM data to rank teams and players and profile playing styles.

Dunk Vision is a desktop basketball shot tracker and data capture tool built using Python and tkinter. Designed for youth coaches, players, and parents, it lets users track, visualize, and analyze shot data to improve individual and team performance as well as direct training efforts. For courtside users, data can be exported for later analysis.

A professional NBA scouting application that combines live player data from the official NBA Stats API with a two-stage AI agent architecture. The system researches player tendencies, identifies comparable players, fetches their live stats, and synthesizes everything into an executive scouting memo.

Database for the BYTHERIM Fictional Recruiting Database with player rankings, class rankings, and a comprehensive statistics page.

Automated NBA shot chart and heat map generator built with Python, R, and the NBA API.

An open-source film room for coaches, clip a game by talking to it. Local, offline, no account.

An Android application designed to help referees and staff coordinate, organize, and manage basketball matches for Club Bob1. The app aims to simplify match planning, staff coordination, game management, and communication before basketball games.

Possibility-Adjusted Shot Value — Companion code, data, and pre-registration for the SSAC27 paper "Every Shot Is a Measurement: A Theory of Possibility Cost in NBA Possession Value"

Multi-scenario analysis of how NBA trash talk and conflicts affect player performance

Claude tool-use agent that writes basketball scouting reports: 7 tools across Basketball Reference, ESPN, EuroLeague and FIBA, 5-call budget per report, 24h Redis cache, wrong-player validation exposed at /metrics

NBA research workbench: versioned Postgres/dbt warehouse, play-by-play RAPM and impact metrics, out-of-time forecast evaluation, and a tool-calling assistant. Python, FastAPI, Next.js.

WNBA lineup stints and player impact ratings, 2003-2026. 267,293 stints reconstructed from play-by-play, with ARC ratings and channel decompositions.

A full CV basketball analytics system powered by YOLOv12, ByteTrack, SuperPoint/LightGlue, FastAPI, MongoDB, minIO, Prometheus and Grafana to turn raw basketball broadcast footage into rich spatial analytics and visualized insights.