Horse Racing Favorites Analysis

I’m sitting on almost four years of RBD race data and want to turn it into clear, actionable insight about favourites. The goal is to quantify how often the market’s top pick wins, loses or merely breaks even, then compare that reality with the odds-implied expectation. I care about the pattern through time—daily, weekly and monthly swings—so I can spot “hot” or “cold” stretches rather than back-fitting a single global figure.

Key filters must be baked in from the start: race type, horse age, track conditions and each runner’s RBD Rank. I need the analysis to respect those dimensions without data-snooping, so whatever modelling approach you take should allow me to hold-out or cross-validate properly.

Because I ultimately want to decide whether long-term betting or laying offers the better edge, please include both win-rate and ROI style metrics in the final output. A basic understanding of book-price maths (over-round, true probability, implied margin) is essential so the results make sense in betting terms.

Deliverables
• Clean, well-commented code (Python, R or similar) that loads the dataset, applies the chosen filters and produces daily/weekly/monthly summaries
• A predictive model or probability calibration that sets the “expected” benchmark for each favourite
• Comparative tables/plots showing actual vs expected win percentage and cumulative ROI for both back and lay strategies
• A concise write-up of methods, assumptions and any sensitivity checks, so I can rerun the study on future data without accidental back-fit

I’ll supply the raw CSVs once we start; everything else—processing, modelling and visualisation—should come back reproducible end-to-end.