The WINDEX Model (Winning Index Model for Draw No Bet) is a statistical framework designed to predict the likelihood of home and away wins in football matches, specifically for the Draw No Bet (DNB) market. The model focuses exclusively on historical performance data and does not rely on bookmaker odds, making it an objective tool for identifying betting opportunities.
Core Components
- Inputs:
- Team Strength Metrics: Home/away win rates, ELO ratings, league rankings.
- Form Indicators: Points per game, goal differences, and recent streaks (e.g., last 5–10 matches).
- Head-to-Head Results: Historical outcomes between the two teams.
- Goals Data: Goals scored and conceded (home/away splits).
- Draw Rates: Team and league-specific draw probabilities.
- Feature Engineering:
- Win probability estimation based on home/away splits and team form.
- Momentum index derived from rolling averages of goals and points.
- Adjustments for draw likelihood to account for DNB refund scenarios.
- Model Process:
- Win Probability Estimation: Use historical data to calculate the probabilities of a home or away win while excluding the draw.
- Refund Likelihood: Integrate draw probabilities to assess the likelihood of refunded bets.
- Outcome Confidence: Provide a confidence level for each team’s win probability.
- Outputs:
- Predicted probabilities for home and away wins.
- Actionable insights to identify value in DNB betting, particularly for close matches.