PuckSnipe - Sell Project

Hey how's it going everyone.I'm currently looking to sell a desktop software that I'm working on called PuckSnipe. PuckSnipe is a full‑stack NHL prop‑betting intelligence app that blends AI matchup analysis with real stats and optional local ML models. Users pick games, run batch analysis, build slips, and track results. The frontend is a polished React + Vite UI; a companion FastAPI service powers a goal‑scorer model, SOG model, and goalie resolver. Demo mode is included for zero‑friction previews. Core Features - AI matchup analysis with a structured “Panda Protocol” prompt that outputs 3–6 prop bets per game, including edge, reasoning, confidence, odds guidance, and goalie context. - Multi‑game batch analysis with progress tracking and clean result grouping. - Prop type filters (SOG, Goal, Assist, Point). - Bet cards with headshots, “last 10” hit‑rate visualization, confidence bar, and ML overlays. - Bet slip builder with drag‑to‑open widget and export to PNG. - My Bets tracker with local storage persistence, filters, sorting, manual result tagging, NHL stat verification helper, and bulk ZIP export of cards. - ML dashboard for uploads, training triggers, model health, feature importance, and metrics. AI / ML + APIs - Perplexity AI (Sonar Pro): Generates the core analysis response for each game. The system prompt enforces bet count, formatting, and decision heuristics. - Odds API (v3): Fetches NHL schedule and Hard Rock Bet player props. - NHL API (proxied through Vite): - search.d3.nhle.com for player ID lookup - api-web.nhle.com for game logs, gamecenter landing data, and schedule‑adjacent data - DailyFaceoff goalie resolver: Scrapes starting goalies (HTTP first, Playwright fallback). - Optional local ML service (FastAPI + XGBoost): - /predict goal‑scorer probability - /predict/skaters/sog expected SOG + over‑line probability - /goalies DailyFaceoff goalie slate - /h2h shooter‑vs‑goalie historical rates - /train and /train/sog model training triggers - /status and /status/sog training state and metrics - /upload and /files CSV ingestion and file history ML Data Sources - MoneyPuck shots data (for goals model) and team skater game‑by‑game logs (for SOG model). - Training outputs stored as JSON artifacts (model file, feature importance, metrics, and training state). Tech Stack- Frontend: React 19, Vite 6, TypeScript, Tailwind (CDN), lucide‑react, html‑to‑image, JSZip. - Backend: Python, FastAPI, XGBoost, Pandas, scikit‑learn, BeautifulSoup, requests, Playwright (optional).- UI components for game list, bet cards, slip, tracking, ML dashboard, feature charts, and last‑5 graphs.Data & Storage- Local storage for API keys and settings (user_perplexity_key, user_odds_key, ml_service_url, ml_service_enabled) and tracked bets (puckprophet_bets).- Optional local datasets in Shots Data/ and skaters_team_gbg/ for ML training.- Local headshot cache under public/players/.Included Scripts & Utilities- Headshot downloader and demo asset checks.- Player ID DB generator.- Goalie resolver debug script.- ML data inspection scripts.It's great already and I've used it a few times with the Hard Rock Bet app. More features could be easily added. Could be used for selling NHL picks or personal use.Everything is a folder and I've been using npm run dev. Haven't turned it into an executable.