Determinants of Scoring at Castle Pines Golf Club

This analysis is based on scores and stats from individual rounds in the five PGA Tour events at Castle Pines Golf Club: 1,595 rounds in total.

Section 1: Absolute Correlation Coefficients with Score

Traditional Metrics Correlation

The graph above illustrates how the absolute value of the correlation coefficient between the players' scores and traditional golf metrics - such as Driving Distance, Driving Accuracy, Greens in Regulation, Scrambling, and PPGIR - varies by year.

Driving Distance: The correlation between Score and Driving Distance is relatively low across the years. This suggests that, at Castle Pines Golf Club, driving distance does not strongly predict overall performance. A possible explanation could be that the course design diminishes the advantages of longer drives, emphasizing accuracy and short game instead.

Driving Accuracy: The correlation here is moderate, implying that players who keep the ball in play (i.e., in the fairway) tend to score better. Castle Pines, with its challenging rough and strategically placed hazards, likely penalises wayward drives, making accuracy more critical than distance.

Greens in Regulation (GIR): This metric shows the highest correlation with score, especially in certain years. This strong relationship indicates that reaching the green in regulation is crucial for a good score. Castle Pines' greens are known for their complexity, so players who can consistently reach the green in regulation are at a distinct advantage.

Scrambling: The correlation between scrambling and score is also quite strong. This highlights the importance of a player's ability to recover from missed greens. Given the challenging nature of Castle Pines, where even slight errors can lead to difficult lies, a strong scrambling game is essential for minimising bogeys and worse.

PPGIR (Putts per Green in Regulation): The moderate to high correlation here underscores the importance of putting. Castle Pines is renowned for its fast, undulating greens, making putting a key determinant of success. Players who can convert their opportunities on the green tend to score better.

Par Metrics Correlation

The second graph depicts how the absolute value of the correlation coefficient between the players' scores and their performance on Par 3s, Par 4s, and Par 5s changes over time.

Par 3 Performance: The correlation between score and Par 3 performance is moderately strong. This indicates that success on Par 3 holes is a significant factor in overall scoring at Castle Pines.

Par 4 Performance: The strongest correlation is found with Par 4 performance. This is unsurprising, as Par 4 holes typically make up the bulk of a golf course's layout.

Par 5 Performance: The correlation between score and Par 5 performance is also significant but tends to be lower than that for Par 4s.

Section 3: Importance of Each Metric in Determining Score

Random Forest Regressor and Feature Importance

Random Forest Regressor is an ensemble learning method that constructs multiple decision trees during training and outputs the average prediction. It combines the predictions of several models to improve accuracy and robustness.

Feature importance is a technique used to interpret a machine learning model. It refers to the score that quantifies the contribution of each feature to the prediction made by the model.

In a Random Forest, the importance of a feature is computed by looking at how much the feature decreases the impurity (e.g., variance for regression tasks) across all the trees in the forest. The more a feature decreases the impurity, the more important it is considered.

The calculated importance scores for all features are then normalized to give relative importance as a percentage. This shows the relative contribution of each feature to the prediction task.

Interpreting Feature Importance

Features with high relative importance percentages have a strong impact on the model's predictions. They are crucial for accurate predictions and indicate key areas where performance matters most.

Features with low relative importance have a minimal impact on the model's predictions. While they can still contribute, they are less critical.

Section 2: Importance of Each Metric

Traditional Metrics Importance

The bar chart visualises the relative importance of various traditional golf metrics - such as Driving Distance, Driving Accuracy, Greens in Regulation, Scrambling, and PPGIR - on the players' scores as determined by a Random Forest Regressor.

Greens in Regulation (GIR): With an importance of approximately 40.32%, GIR is the most significant predictor of score. This highlights the critical role of consistent approach shots in reaching the green in the expected number of strokes, particularly at Castle Pines Golf Club, where the greens are known to be challenging.

Scrambling: Contributing 29.09% to the model, scrambling is also a crucial metric. This further emphasizes the importance of a player's ability to save par after missing the green, which is often necessary on this tough course.

PPGIR (Putts per Green in Regulation): PPGIR accounts for 20.71% of the model's predictive power, indicating that while putting is important, its impact is somewhat secondary to GIR and scrambling at Castle Pines.

Driving Distance: At 6.51%, Driving Distance plays a relatively minor role in determining score. This suggests that the course design at Castle Pines may reduce the advantage of longer drives, possibly due to tight fairways or strategic hazards.

Driving Accuracy: The least impactful metric is Driving Accuracy, with an importance of 3.37%. While accuracy off the tee is always beneficial, its lower relative importance may indicate that players can recover well from missed fairways or that the course does not penalize inaccuracies as severely as others might.

Summary: The findings at Castle Pines Golf Club differ from the average PGA Tour statistics. Notably, the importance of Greens in Regulation (40.32%) is higher than the PGA Tour average (29.77%). Scrambling (29.09%) and PPGIR (20.71%) are also vital but slightly less critical compared to their PGA Tour averages of 27.02% and 30.13%, respectively. Driving Distance and Accuracy are less significant at Castle Pines than on the average PGA Tour course, where they account for 9.31% and 3.77%, respectively.

Par Metrics Importance

The bar chart shows the relative importance of Par 3, Par 4, and Par 5 performance on players' scores as determined by a Random Forest Regressor.

Par 4 Performance: With an overwhelming importance of approximately 67.03%, Par 4 performance is the most significant predictor of score. This aligns with expectations, as Par 4 holes typically make up the majority of a golf course, and success on these holes often requires a well-rounded skill set.

Par 5 Performance: Par 5 performance contributes 21.37% to the model, indicating that while Par 5s are opportunities for birdies or better, they are less critical than Par 4s in determining the overall score.

Par 3 Performance: At 11.60%, Par 3 performance is the least significant factor. Par 3s are typically shorter and demand accuracy and precision, but they represent a smaller proportion of the total holes on the course than Par 4s.

Summary: The findings at Castle Pines Golf Club show a strong emphasis on Par 4 performance, consistent with the average PGA Tour statistics but with some variations. The importance of Par 4 performance (67.03%) is very close to the PGA Tour average (67.12%). Par 5 performance (21.37%) is slightly more important than the PGA Tour average (15.56%), possibly due to the specific design and scoring opportunities provided by Par 5s at Castle Pines. Par 3 performance (11.60%) is somewhat less important compared to the PGA Tour average (17.32%), suggesting that while precision is important, it has a smaller impact on the overall score relative to the longer holes.

Top 5 Ranked Players - 2024 BMW Championship

The table below shows the top-5 ranked players and their average estimated scores from the different Random Forest models above.

Player Score
Scottie Scheffler 68.06
Xander Schauffele 68.56
Ludvig Aberg 68.79
Davis Thompson 69.32
Aaron Rai 69.88

Estimated scores for all players can be found here.