Provides tools for querying and analyzing a local DuckDB warehouse containing NFL play-by-play data, player/team stats, rosters, injuries, officials, draft history, and betting lines, enabling AI agents to answer football questions via SQL.
Click on "Install Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@NFL Analytics MCP which QBs perform best traveling east?"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
A personal NFL analytics platform that runs entirely on your machine: a local warehouse of every NFL play since 2007, a visual dashboard, auto-updating data and news, a prediction-market price tracker, and an AI analyst you can ask questions in plain English.
No API keys. No subscriptions for the core experience. One ~2 GB download.
Piece | What it does |
Warehouse (DuckDB) | 909k+ plays (2007–2025), player/team stats, rosters, injuries, officials, draft history, and the full 2026 schedule with betting lines — all queryable in milliseconds |
Jarvis UI (React + FastAPI) | A dark, glowing "command center": division constellation with all 32 team logos → per-team HUDs in team colors (stat rings, efficiency charts, roster, coach lineage) → live prediction-market board → built-in streaming AI chat (Ctrl-K) |
Dashboard (Streamlit) | The simpler original UI: division standings → team pages → players, league leaders, schedules & lines |
News engine | Auto-polls ESPN plus all 32 official team websites every 6 hours; headlines are tagged to players/teams in the warehouse |
Kalshi tracker | Records prediction-market prices (game winners, spreads, totals, win totals, Super Bowl futures) every 6 hours, building line-movement history |
Prediction model | Opponent-adjusted EPA ratings → win probabilities, honestly backtested against 18 years of closing lines (spoiler: Vegas wins — the model's value is calibration, and the report shows exactly by how much) |
AI analyst | A chat page (and MCP server for Claude Code/Desktop) that writes and runs real SQL against your warehouse to answer questions like "which QBs perform best traveling east?" |
Related MCP server: Feyod MCP Server
See SETUP.md for the full guide. The short version:
git clone https://github.com/parthakker/nfl-analytics.git
cd nfl-analytics
pip install -e .
python scripts/refresh_data.py --bootstrap # ~2 GB from nflverse, one time
python -m streamlit run dashboard.pynflverse releases ─┐ (nightly-updated public data)
ESPN + team sites ─┼─► scripts/refresh_data.py / poll_news.py / snapshot_kalshi.py
Kalshi API ────────┘ │ (scheduled: weekly / 6h / 6h)
▼
nfl.duckdb + news.duckdb + kalshi.duckdb
│
┌─────────────────┼──────────────────┐
▼ ▼ ▼
dashboard.py MCP server (15 tools) model/
(Streamlit UI) (Claude Code/Desktop) (ratings, backtest)Design principles: compute, don't retrieve (questions are answered by SQL
over plays, not by searching documents); verified semantic layer (every
data gotcha — and NFL data has many — is documented in docs/dictionary/ and
enforced in CLAUDE.md); honest modeling (walk-forward backtests with an
untouched holdout, reported even when the answer is "the market is better").
All stats data from the outstanding nflverse project. News from ESPN and official team site feeds. Market data from Kalshi's public API. This is a personal, non-commercial project; all data remains property of its respective owners.
MIT — see LICENSE.
Unclaimed servers have limited discoverability.
If you are the server author, to access and configure the admin panel.
Query PostgreSQL databases in plain English — LLM-generated, safety-validated SQL.
GibsonAI MCP server: manage your databases with natural language
The grounded data layer for any LLM: governed SQL, metrics, lineage and catalog over your data.
We provide all the information about MCP servers via our MCP API.
curl -X GET 'https://glama.ai/api/mcp/v1/servers/parthakker/nfl-analytics'
If you have feedback or need assistance with the MCP directory API, please join our Discord server