Player Impacts Detection

The player impact metric algorithm is designed to quantify impacts across various types of collisions during gameplay, including tackles, being tackled, blocks, and creating contact with the ground. Multiple impacts can occur on the same possession if multiple collisions happen outside the detection window of an impact.

Development Process

Data Collection: Player data across multiple offensive and defensive positions from NCAA Division I players was collected using Catapult S5 devices. This data included information on movements, collisions, and player loads.

Machine Learning Model Training: A machine learning model was trained on the collected player data to recognize patterns and characteristics associated with different types of impacts. The model was trained to accurately identify and classify impacts based on the sensor data.

Testing and Validation: The algorithm was tested on multiple positions (excluding linemen) to ensure its accuracy and reliability. Testing yielded targeted results above 85% accuracy, indicating the algorithm’s effectiveness in detecting impacts.

Output and Use Case

The primary use case of this detection algorithm is to output total impacts and the associated PlayerLoad for further training and game-based analysis. The algorithm provides valuable insights into the physical demands placed on players during gameplay, allowing coaches, sports scientists, and training staff to monitor and analyze player performance, manage workload, and optimize training strategies.

By accurately quantifying player impacts, this metric contributes to a comprehensive understanding of player performance and injury risk, enabling data-driven decision-making and player management strategies.