Soccer Oracle uses XGBoost and AI to predict football match outcomes across 10+ leagues — Premier League, Bundesliga, La Liga, Serie A, Ligue 1 and more. Each match shows win/draw/loss probabilities, over/under odds (1.5–4.5), and an AI breakdown with H2H history and recent form. The Ticket Generator filters today's matches by market and confidence threshold, returning the strongest picks sorted by probability. No sign-up. No paywall. Just data.

Wispr Flow: Dictation That Works Everywhere — Stop typing. Start speaking. 4x faster.

Stop typing. Start speaking. 4x faster.

I started building Soccer Oracle because I was tired of "prediction" sites that give you a result with no reasoning behind it — just someone's opinion dressed up as analysis. The core is an XGBoost model trained on years of historical match data and league standings. On top of that, OpenAI generates a written pre-match breakdown for each fixture — covering H2H history, recent form, and key contextual factors. The hardest problem was surprisingly mundane: team name normalization. ESPN calls a team "Cologne", my database has "1. FC Köln". Getting fuzzy matching right across 5 languages took longer than the model itself. The History page shows predicted vs actual results — I wanted full transparency on where the model is right and where it fails. Would love feedback on the UX and on what features would actually make this useful for you.

Wispr Flow: Dictation That Works Everywhere — Stop typing. Start speaking. 4x faster.

Stop typing. Start speaking. 4x faster.

Soccer Oracle

Win/draw/loss probabilities for every match, powered by ML

Win/draw/loss probabilities for every match, powered by ML