Betting Oracle — Data Analysis for Sports Betting

Betting Oracle is a compact hub that collects and showcases my sports‑prediction projects and dashboards in one place. Each card below links to a repository or app and includes a short summary so you can quickly explore the model, data sources, and how to run the project. Click an icon to open the repo or follow the quick‑start steps in each README to run the Streamlit apps locally.

Equine Edge - Horse Racing Predictions
(www.equine-edge.horse)

A comprehensive ML system for UK horse-racing with a Streamlit dashboard, data pipeline, and XGBoost win prediction model. Includes a race profitability scorer, fixture calendar with predicted scores, and utilities to convert probabilities into betting odds.

The system identifies the most profitable races to bet on using a smart scoring algorithm that factors in class quality, prize money, course tier, and field size. The XGBoost model predicts win probabilities with 18 carefully engineered features, and the dashboard lets you filter by year, course, and horse to explore patterns and compare model odds against bookmaker odds for value bets.

Pitch Oracle - Premier League Predictor
(www.pitch-oracle.com)

A Premier League match predictor with a Streamlit UI and ensemble modeling (XGBoost, Random Forest, Gradient Boosting, Logistic Regression) plus optional neural/LSTM support. Shows upcoming fixtures, predicted probabilities, referee/manager statistics, and model-comparison metrics with scripts to fetch and process fixtures.

This project goes deeper than just match outcomes—it analyzes referee tendencies, manager performance trends, and how recent team form impacts results. The ensemble approach combines multiple algorithms and achieves 3.5% higher accuracy than using XGBoost alone. You can explore historical data, see detailed statistics by referee or manager, and understand the confidence level behind each prediction.

Gridiron Oracle - NFL Predictions
(www.gridiron-oracle.com)

A sophisticated NFL analytics platform with multi-model ML pipelines and a multi-page Streamlit dashboard. Offers player props, DraftKings Pick 6 tools, automated nightly prediction updates, and production-ready features for generating and exporting betting insights.

This is one of the most feature-rich systems—it can predict spreads, moneylines, and player prop performances with separate specialized models. The dashboard includes a bankroll management tool to calculate optimal bet sizes, an in-app notification system to alert you about high-confidence opportunities, and automated nightly updates that regenerate predictions using the latest game data and betting lines. There's even a per-game detail view with shareable links and historical play-by-play analysis.

Gridlocked Oracle - F1 Analysis
(www.gridlocked.racing)

Formula 1 data analysis tools and a Streamlit app that generate race analytics from F1DB and FastF1 sources. Includes generators for CSV/JSON datasets, multiple model support, email notifications, and extensive feature engineering for predictive modeling.

With 86+ engineered features and support for four different ML algorithms (XGBoost, LightGBM, CatBoost, Ensemble), this system achieves a Mean Absolute Error of 1.94 for predicting race finishes. It pulls data from F1DB's massive historical dataset plus real-time race control messages from FastF1, and includes advanced filtering for over 30 different parameters. Email notifications deliver pre-race predictions with team colors and driver rankings, making it easy to stay updated on every Grand Prix.

Bracket Oracle - March Madness Predictor
(www.bracket-oracle.com)

A March Madness betting prediction system that integrates historical data, KenPom and BartTorvik efficiency ratings, and ML models. Provides real-time predictions, underdog value detection, Kelly Criterion sizing, and automated data pipelines for tournament analytics.

This system combines a decade of tournament data with advanced efficiency metrics from KenPom and BartTorvik to improve spread predictions by 8.4%. It automatically detects underdog value bets where the model disagrees with the sportsbook odds, calculates optimal bet sizing using Kelly Criterion, and continuously updates as new team data arrives. The NCAA selection is populated in real time, and the system automatically canonicalizes team names across all data sources so you get consistent, clean predictions.

Oracle on Ice - NHL Predictor
(www.oracle-on-ice.com)

A Streamlit-powered analytics platform designed to provide data-driven insights for NHL sports betting. This application aggregates real-time and historical NHL data to help bettors make informed decisions.

The app pulls comprehensive statistics from NHL APIs, identifies betting value by surfacing statistical edges, visualizes patterns in historical performance and trends, and tracks prediction accuracy over time. Features include today's games with real-time odds, team stats, standings, value finder for betting opportunities, player props analysis, and performance tracking.

Fairway Oracle - Golf Predictor
(www.fairway-oracle.com)

Fairway Oracle is a research-focused PGA tournament winner predictor built with Streamlit. It trains ML models on multiple years of historical data to produce top-N winner probabilities, player analytics (strokes-gained, form, course history), and identifies value bets by comparing model probabilities with market odds.

Features include backtesting, live tracking during events, automated scraping of free public sources (PGA Tour, ESPN, OWGR, Wikipedia, weather), and simple deployment via streamlit. Not financial advice — intended for analytics and research.

Breakpoint Oracle - Tennis Predictor
(www.breakpoint-oracle.com)

Breakpoint Oracle is a comprehensive tennis match‑prediction system with a Streamlit interface powered by historical data (TennisMyLife, tennis-data.co.uk) and live pre‑match odds via Matchstat/RapidAPI. A full feature engineering pipeline generates ELO ratings, serve/return statistics, surface‑specific form, head‑to‑head counts, and market probabilities; a trained ML model produces top‑N winner probabilities and highlights betting edge when model confidence exceeds market odds.

The app includes three tabs—Today’s Matches (live odds & edge), Match Explorer (filterable historical dataset), and ELO Rankings—and supports backtesting, automated daily updates through GitHub Actions, and easy local or Streamlit‑Cloud deployment. Not financial advice; intended for analysis and research.

Betting Cleanup - Baseball Predictor
(www.betting-cleanup.com)

Betting Cleanup is a powerful baseball match‑prediction system now maintained in its own GitHub repo. It features a Streamlit interface powered by up‑to‑date historical data (Baseball-Reference, FanGraphs) along with live pre‑match odds pulled from Matchstat/RapidAPI. The full data pipeline engineers ELO ratings, batting/pitching analytics, surface‑specific form, head‑to‑head records, and implied market probabilities; an ML model then generates top‑N winner probabilities and flags potential betting edges when model confidence outpaces market lines.

Recent updates add a fourth dashboard tab—Live Tracker—for real‑time in‑game odds, plus improved backtesting, automated nightly updates via GitHub Actions, and simplified deployment on Streamlit Cloud or locally. Always for research and analysis; not financial advice.

Betting Baseline - NBA Predictor
(www.betting-baseline.com)

Betting Baseline is a Streamlit app for NBA game and DraftKings Pick 6 predictions, built on a robust pipeline that pulls NBA data from nba_api, Basketball Reference, and market lines from The Odds API. It offers game-level forecasts (win probability, spread, total), team/player trends, and player prop probability models with value-edge detection versus sportsbook odds.

The platform includes injury and lineup monitoring, referee assignment cleanup, daily automated data refresh via GitHub Actions, model calibration tracking, and historical backtesting reports. It is designed for analytics research and transparent model evaluation rather than betting advice.