For each match: the chances of a home win, a draw and an away win from the two teams' ratings now, by the projection the projected finish is built from; the goals it expects each side to score; and the chance of more than 2.5 goals in the match. Beside them, the bookmakers' forecast now (Bet365, from football-data.co.uk's upcoming fixtures, the bookmaker's margin taken out).
At lineup time: once the starting elevens are announced, about an hour before kickoff, a second forecast is made from the same projection plus two facts about each eleven, the average rating of the players starting and how many of the team's regular players are not starting. On the other fourteen leagues' seasons 2019/20 to 2022/23 it improved the forecast by 0.0027 in log loss per match. The week-ahead forecast above stays the forecast of record; the lineup-time one is shown beside it and scored on the forecasts page. "Regulars out" is the count of each side's usual eleven not starting.
| Kickoff | Home | Away | Home win | Draw | Away win | Expected goals | Over 2.5 goals | Bookmakers: home | draw | away | At lineup time: home | draw | away | Regulars out |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2026-10-17 11:30 | Everton | Chelsea | 33% | 24% | 43% | 1.3 – 1.5 | 55% | 29% | 25% | 46% | 38% | 25% | 37% | 4 / 7 |
| 2026-10-17 14:00 | Brentford | Liverpool | 33% | 23% | 43% | 1.4 – 1.6 | 59% | 42% | 25% | 33% | ||||
| 2026-10-17 14:00 | Fulham | Hull | 64% | 20% | 16% | 2.0 – 1.0 | 57% | 56% | 24% | 20% | ||||
| 2026-10-17 14:00 | Man City | Ipswich | 78% | 14% | 8% | 2.5 – 0.9 | 67% | 81% | 12% | 7% | ||||
| 2026-10-17 16:30 | Newcastle | Aston Villa | 43% | 24% | 33% | 1.6 – 1.4 | 58% | 43% | 26% | 31% | ||||
| 2026-10-18 13:00 | Bournemouth | Sunderland | 52% | 24% | 24% | 1.7 – 1.1 | 53% | 47% | 27% | 26% | ||||
| 2026-10-18 13:00 | Brighton | Crystal Palace | 52% | 23% | 25% | 1.7 – 1.2 | 57% | 64% | 20% | 16% | ||||
| 2026-10-18 13:00 | Leeds | Man Utd | 32% | 24% | 44% | 1.3 – 1.5 | 54% | 41% | 26% | 33% | ||||
| 2026-10-18 15:30 | Nottingham Forest | Arsenal | 21% | 23% | 57% | 1.0 – 1.7 | 53% | 15% | 24% | 61% | ||||
| 2026-10-19 19:00 | Tottenham | Coventry | 51% | 23% | 25% | 1.7 – 1.2 | 55% | 67% | 20% | 13% | ||||
This round: expect about 4.6 home wins, 2.2 draws and 3.2 away wins, the sums of the matches' chances. The chances are the forecast; they say how many of each outcome to expect across the round, not which matches will produce them.
Every match has a real chance of a draw, usually 25 to 30 per cent, so about one match in four ends level. But in no single match is the draw the biggest of the three chances: it is a strong second or third, never first. Ten matches each given 45 per cent home, 27 draw and 28 away would produce about four or five home wins, three draws and three away wins, while a forecast naming the biggest chance would say "home win" ten times. That is why this page shows the chances and the round's expected counts, not one outcome per match.
A forecast that named one outcome per match, the one with the highest chance, would never name a draw: across the 1,514 Premier League matches of 2019/20 to 2022/23, 23 per cent of which were drawn, the draw had the highest chance in none. For the draw to be the most likely outcome it needs more than a third of the chance, and that happens only when two things hold at once: the teams are nearly equal, and both are expected to score few goals. Between two equal teams the projection gives the draw 39 per cent when each is expected to score 0.7 goals, 34 per cent at 0.9, 32 per cent at 1.0 and 26 per cent at 1.4. Premier League sides are expected to score 1.2 to 1.5 goals a match, and the home side's advantage makes most matches unequal, so the draw is never on top. In the 107 matches between nearly equal teams the chances averaged 37 per cent for the home win, 25 for the draw and 37 for the away win, and the draw's chance never passed 30. The bookmakers agree: Pinnacle's closing forecast never made the draw the most likely outcome in those 1,514 matches either; its highest was 34 per cent.