Traditional tennis statistics, such as aces, double faults, winners, and unforced errors, provide intuitive box-score summaries of player performance. However, these metrics evaluate isolated aspects of performance rather than summarizing a player's overall performance in a single measure. As a result, comparing individual players often requires tables, radar charts, or other visualizations that present multiple metrics together. This makes interpretation less direct.
Composite metrics address this limitation by combining multiple aspects of play into a single numerical measure. For example, metrics such as Dominance Ratio (DR) and hold-plus-break percentage summarize performance more directly by combining serve and return outcomes. However, these metrics are deterministic summaries of observed outcomes and do not explicitly account for how the value of a point outcome varies with score context. Therefore, they do not distinguish, for example, a winner on break point from a winner at 40–0, even though the two outcomes can have different effects on the probability of winning the current game.
This project introduces Event-Driven Gain in Expectancy (EDGE), a probabilistically derived composite metric for player performance. Using tour-wide point-by-point match data, EDGE empirically quantifies the value of positive and negative point-level event types, such as aces, winners, double faults, and unforced errors, based on their average impact on the probability of winning the current game. A player's EDGE is then calculated as a linear weighted sum of event counts, normalized by the total number of points played. Higher EDGE indicates that a player produces a more valuable overall mix of positive and negative events, based on their tour-wide average effects on the probability of winning the current game.
Log in or sign up for Devpost to join the conversation.