An updated assessment of the predictive accuracy of World Tennis Number and Universal Tennis Ratings
Authors:
- Lucas KielyLKLucas Kiely
- Rebecca MayewRMRebecca Mayew
- William MayewWMWilliam Mayew
Open Access
- 16 December 2025
- journal article
- Published by World Tennis in ITF Coaching & Sport Science Review
- Vol. 33 (97) , 53-61
Abstract
In 2024, the World Tennis Number (WTN) algorithm was updated to enhance its predictive accuracy. This study analyzes 4,174 matches involving elite-level junior development players in the 12-and-under (12U) and 14-and-under (14U) categories, as well as Intercollegiate Division 1 players, from tournaments sanctioned by the United States Tennis Association (USTA), United Kingdom Lawn Tennis Association (LTA), Tennis Europe (TE), and the Intercollegiate Tennis Association (ITA). We compare the updated WTN and Universal Tennis Rating (UTR) classification accuracy in predicting match winners across these competitions. Our findings reveal no consistent evidence that WTN outperforms UTR in any category. UTR exhibits superior predictive accuracy in junior tournaments and in prequalifying rounds at the collegiate level, which highlights rating differences when players transition into new competitive environments. This study provides new evidence on the predictive accuracy of WTN and UTR in the context of young elite junior players and top collegiate talent.Keywords
This publication has 6 references indexed in Scilit:
- Initial evidence on the impact of the 2023 World Tennis Number algorithm change for predicting match outcomesITF Coaching & Sport Science Review, 2025
- World Tennis Number: The new gold standard, or a failure?ITF Coaching & Sport Science Review, 2023
- Age and sex-related upper body performance differences in competitive young tennis playersPLOS ONE, 2019
- Are differences in ranks good predictors for Grand Slam tennis matches?International Journal of Forecasting, 2010
- Are sports seedings good predictors?: an evaluationInternational Journal of Forecasting, 1999
- Comparing the Areas under Two or More Correlated Receiver Operating Characteristic Curves: A Nonparametric ApproachBiometrics, 1988