ATP predictions 9 September 2026
ATP matches only: win probability, most likely set score and deciding-set chance for every singles match, from ratings built on 57,590 matches.
6Matches
1Tournaments
3Forecast
9,192Players rated
Strongest calls of the day
23:30US Open, MenR27Alexander ZverevBotic Van de Zandschulp3–0AI prediction89%High confidence11%ForecastZverev to win89%Full analysis 03:05US Open, MenR27Ben Shelton3Carlos Alcaraz21–3AI prediction36%Medium confidence64%ForecastAlcaraz to win64%Full analysis 21:05US Open, MenR27Karen Khachanov2Alexander Blockx00–2AI prediction40%Medium confidence60%ForecastBlockx to win60%Full analysis
All matches by tournament
ATP US Open, Men
03:05Ben SheltonCarlos Alcaraz
66613
71362
Alcaraz64%1–3›21:05Karen KhachanovAlexander Blockx6732
2520
Blockx60%0–2›22:00Krawietz K / Puetz TArevalo M / Pavic M
6672
7461
Doubles22:10Ram R / Salisbury JGranollers M / Zeballos H
662
440
Doubles23:10Heliovaara H / Patten HAndreozzi G / Guinard M
23:30Alexander ZverevBotic Van de Zandschulp6762
7621
Doubles–
–
Zverev89%3–0›
How these predictions are made
Every player carries a rating that moves after each match by how surprising the result was — beating a stronger opponent is worth more than beating a weaker one, and a win in an ATP or WTA draw counts for more than one in an ITF event. The win probability comes from the gap between two ratings, adjusted by their head-to-head record. The set score follows from that probability rather than being guessed separately. Where a player has played too few matches, or has never played at tour level, the rating is pulled back towards the average and the match is marked as thin data instead of being dressed up as a confident call. Doubles are excluded: the rating measures singles.
Every player carries a rating that moves after each match by how surprising the result was — beating a stronger opponent is worth more than beating a weaker one, and a win in an ATP or WTA draw counts for more than one in an ITF event. The win probability comes from the gap between two ratings, adjusted by their head-to-head record. The set score follows from that probability rather than being guessed separately. Where a player has played too few matches, or has never played at tour level, the rating is pulled back towards the average and the match is marked as thin data instead of being dressed up as a confident call. Doubles are excluded: the rating measures singles.