1. Abstract

Arrigo is Driblab’s last innovative product that integrates event and tracking data to deliver a comprehensive set of metrics. Designed to support data-driven decision-making, Arrigo provides detailed insights into player actions, team dynamics, and strategic execution on the field. It delivers a wide range of metrics focused on critical areas of play, from bypassing opponents to tracking off-ball movements. With this tool, teams can analyze how their strategies unfold in real time, identify performance patterns, and make data-driven adjustments to improve outcomes on the pitch.

2. Match Preprocessing

2.1.1 Data Retrieval

Tracking Data:

• We use the Driblab API to obtain the AWS link that allows us to download the tracking JSONL file.

Event Data:

• We use the Wyscout API to retrieve the events that occur during the game.

Offsets

• We use two different pairs of offsets:
  1. Footage Offsets: These are derived from the footage where ball and player detections were captured.
  2. Event Offsets: These are from Wyscout and align with the events data.
• Using these offsets, we calculate the Offset Diff, which represents the difference between both offsets in the various halves of the game (including extra-time first and second half).
• We later use this difference to synchronize the tracking detections with the events data.

Data Cleaning and Preparation

• We extract and remove the first line of the tracking data, as it contains the plyr_lookup.json.
• We also create the following pickle files:
  • wy_id2plyr_id.pkl: This stores a mapping of the players' Wyscout ID to Driblab's player ID. It's used consistently throughout the process to analyze events and tracking positions.
  • wy_id2team_id.pkl: Similar to the above, but for team IDs.
• Additionally, we create a file called match_info.json, which stores data from the database such as:
  • side_attack
  • home_color
  • away_color
  • Aliases
  • Other relevant information