This role involves the development and maintenance of a proprietary predictive modeling pipeline for Australian and Hong Kong Horse Racing. The objective is to generate accurate win/place probabilities and identify "value" by comparing model outputs against live market prices (Betfair/TAB). Key Responsibilities

• Feature Engineering (The Core): Develop and maintain complex features including sectional time normalization, weight-for-age (WFA) adjustments, and "Beyer-style" speed figures adapted for Australian tracks.
• Predictive Modeling: Train and tune ensemble models (e.g., XGBoost, LightGBM) or deep learning architectures to output calibrated probabilities rather than simple classifications.
• Leakage Control: Rigorously manage temporal data splits to prevent "Look-Ahead Bias" (ensuring the model never trains on data that wouldn't have been available at jump time). 
• Automated Pipeline: Manage the end-to-end flow from data scraping/ingestion (API or web-based) to real-time feature generation and automated bet execution.
• Backtesting & Evaluation: Build and run "Walk-Forward" backtests to validate the Expected Value (+EV) of the betting strategy against historical closing prices. Required Technical Skill Set
• ML Foundations: Strong grasp of Log-Loss optimization, probability calibration (Platt scaling/Isotonic regression), and handling imbalanced datasets.
• Data Engineering: Proficiency in SQL and Pandas for handling "messy" historical racing data (non-runners, track upgrades, late scratchings).
• Domain Knowledge: Understanding of racing-specific variables: barrier bias, jockey/trainer form cycles, and track condition (Firm 1 to Heavy 10) impacts. 
• API Integration: Experience with the Betfair API or similar for real-time price monitoring and order placement.

Full handover available.