The model’s week: 28 September – 4 October
The model’s most likely outcome came true in 152 of 248 matches (61%), against the 63% it expected — well within a normal week’s variation.
- matches predicted
- 248
- picks came true
- 61%
- the model expected
- 63%
- upsets
- 19
By sport
| Sport | Matches | Came true | Expected |
|---|---|---|---|
| Football | 65 | 55% | 54% |
| Hockey | 35 | 57% | 59% |
| Tennis | 148 | 65% | 68% |
Confident calls that came true
Biggest misses
There were 19 upsets — wins by a side the model gave 30% or less — and it expected about 22.6.
Rybakina E.1:2Charaeva A.
6%for the winner94%favourite — Rybakina E.
Potapova A.0:0Kraus S.
13%for the winner87%favourite — Potapova A.
Bartunkova N.2:0Sabalenka A.
13%for the winner87%favourite — Sabalenka A.
Wales2:1Norway
14%for the winner63%favourite — Norway
Lys E.0:1Sun X.
18%for the winner82%favourite — Lys E.
Sun X.2:0Bucsa C.
19%for the winner81%favourite — Bucsa C.
By league
| League | Matches | Came true | Expected |
|---|---|---|---|
| WTA | 83 | 65% | 68% |
| ATP | 65 | 65% | 68% |
| UEFA Nations League | 52 | 58% | 55% |
| NHL | 35 | 57% | 59% |
| Segunda Division | 10 | 30% | 47% |
Only leagues with at least 8 matches in the week: on fewer the share says nothing.
Each match counts with the model’s last estimate before kick-off, the one its page showed; hockey includes overtime. A week runs Monday to Sunday, Moscow time.