Data Preparation
The data was structured in Excel, with reference formulas that allow for quick updates when adding new players. It was then imported into R for analysis. One key feature added was a "LastWeekPoints" column to track each player's fantasy performance from the previous game.
Modeling Approach
A linear regression model was applied to predict fantasy points for Jalen Hurts based on multiple factors:
- LastWeekPoints (Fantasy points in the previous game) - IsHome (Home vs. Away game)- OddsTotal (Total points line from sportsbooks)- Surface (Grass/Turf)-Pass Run Index (PRI) (Offensive balance)- Dome (Indoor/Outdoor stadium)
- LastWeekPoints (Fantasy points in the previous game) - IsHome (Home vs. Away game)- OddsTotal (Total points line from sportsbooks)- Surface (Grass/Turf)-Pass Run Index (PRI) (Offensive balance)- Dome (Indoor/Outdoor stadium)
Results & Observations
- Quarterbacks (QBs): The model predicted fantasy performance within a 6-12% variance using the confidence interval.- Wide Receivers (WRs): Predictions followed a similar accuracy range.- Running Backs (RBs): The model underperformed, with predictions off by 20-30%, suggesting a need for additional variables (e.g., rush attempts, game script).Betting Outcome: Actual wagers based on this model resulted in a 50% loss, indicating that while historical fantasy points provide some insight, they are insufficient as a sole predictor for betting strategies.
Key Takeaways & Future Improvements
1. QB & WR performance can be moderately predicted using historical fantasy data and game-specific variables.2. RB performance is harder to predict using this model, likely due to volume volatility and game script unpredictability.3. Betting applications need refinement—integrating defensive matchups, weather conditions, and advanced metrics may improve predictive accuracy.4. Future iterations could incorporate machine learning techniques (e.g., Random Forest, XGBoost) to capture nonlinear relationships.