Tennis Glossary
The metrics and terms behind Smashrs predictions: how the Elo model rates players, what a player archetype means, and how we turn probabilities into edges against the market.
Model & Metrics
Backtest & Track Record
Replaying the model against historical matches it never trained on, to measure real out-of-sample accuracy and ROI.
Elo Rating
A single number that estimates a player's current strength from match results, and the backbone of the Smashrs model.
Head-to-Head (H2H)
The all-time record between two players, used as a model input and a quick read on a specific matchup.
Model Consensus
How many of the six submodels in the Smashrs ensemble agree on a pick, a confidence gate that varies by tour.
Model Edge
The gap between the model's win probability and the book's implied probability, and how Smashrs spots a mispriced match.
Player Archetypes
Style labels the model assigns from a player's stats (big servers, grinders, all-courters) to capture matchups Elo alone misses.
Surface Elo
Separate Elo ratings for hard, clay, and grass, because the same player is a different player on each surface.