Saturday 31 October, 12:30 UK time
ChelseavManchester United
Pass
No recommendation for this match. The strongest option barely differs from what happens in a typical Premier League match. Strongest candidate, not recommended: Chelsea win or draw (68%).
The candidate would need: Chelsea must avoid defeat: a win or a draw both count.
How likely: the model rates it about 7 in 10 (68%), so in similar matches it would still fail about 3 times in 10. A probability, not a promise.
Why it was considered: The rating model gives “Chelsea win or draw” a 68% chance, against 68% in a typical Premier League match over the last three seasons.
What could spoil it: the likeliest losing score is 1–2 (a model illustration, 7%); together the losing scores make up 32%.
Settled on the final score after 90 minutes plus stoppage time. Postponed, abandoned or cancelled matches are void.
Why this call
The rating model gives “Chelsea win or draw” a 68% chance, against 68% in a typical Premier League match over the last three seasons.
This season Chelsea have scored 17 and conceded 15 in 8 league games; Man United 10 and 13 in 8.
Out-of-sample check (2024/25–2026/27, 234 matches): when the model rated this market 60–70%, it landed 159 times (68%).
What could spoil it
The model still leaves 32% for this to fail; the likeliest losing score is 1–2.
The evidence behind the numbers
What the locked forecast says
Stored when the prediction was locked (30 Oct, 23:10 GMT, version 1, model tl-poisson-v1.1). These numbers are never recomputed, so later results or data cannot change them.
| Result | Probability |
|---|---|
| Chelsea to win | 46% |
| Draw | 23% |
| Man United to win | 32% |
Model scoring expectation: Chelsea 1.79, Man United 1.46 goals. This comes from the team-strength ratings, not from a shot-quality (xG) data feed. Uncertainty ranges and a factor-by-factor breakdown were not stored with this prediction, so none is shown for it. How the live model was tested.
Every possible score Model illustration from the locked scoring expectation; hover or tap a score.
Final score
Chelsea 2–0 Man United
6.2% chance of exactly this score (before kickoff).
Super Tip, Chelsea win or draw: wins
✓ Wins: Chelsea to win · Chelsea win or draw · Under 2.5 total goals · Under 3.5 total goals · Over 1.5 total goals · At least one team fails to score · Chelsea over 1.5 goals · Man United not to score
✗ Loses: Draw · Man United to win · Man United win or draw · Over 2.5 total goals · Both teams to score · Man United over 1.5 goals · Chelsea not to score
Percentages, darker = likelier. ✓ marks scores where the Super Tip wins: together 68.5%, matching the locked 68.5%. Most likely single score: 1–1 (10.2%). Scores with 6+ goals for a side: 1.4%. FT marks the final score.
Every market we priced Compared with a historical base rate, not a bookmaker price.
| Market | Model | Historical base rate | Comparison |
|---|---|---|---|
| Both teams to score | 64% | 58% | |
| Over 2.5 total goals | 63% | 58% | |
| Chelsea over 1.5 goals | 53% | 49% | |
| Man United over 1.5 goals | 43% | 39% | |
| Chelsea to win | 46% | 43% | |
| Over 1.5 total goals | 83% | 82% | |
| Chelsea win or draw ★ | 68% | 68% | |
| Man United to win | 32% | 32% | |
| Man United not to score | 23% | 25% | |
| Draw | 23% | 25% | |
| Man United win or draw | 54% | 57% | |
| Under 2.5 total goals | 37% | 42% | |
| Chelsea not to score | 17% | 22% | |
| Under 3.5 total goals | 59% | 65% | |
| At least one team fails to score | 36% | 42% |
The base rate is how often each outcome happened in Premier League matches over about three seasons; the gap to it is what makes a fixture distinctive. It is not a bookmaker price and says nothing about betting value. Model scoring expectation (from team ratings, not a shot-quality xG feed): Chelsea 1.79, Man United 1.46. Totals and both-teams-to-score markets are priced but not eligible as Super Tips: in testing the model did not beat the base rate there.
Context and data gaps Team news, rest and weather; none of it changes the probability.
No experimental snapshot or context snapshot was stored before kickoff for this match. We do not reconstruct one afterwards, because results and reports collected later would leak into it.
Experimental views Not official calls and never scored. Only the call above is official.
No experimental snapshot or context snapshot was stored before kickoff for this match. We do not reconstruct one afterwards, because results and reports collected later would leak into it.
Low-priority patterns: kickoff time and head-to-head Not used by the model; small samples, mostly explained by team strength.
Chelsea at 12:30 (home)
History since 2020/21Small sample: 8Chelsea won 5 of 8 home league games kicking off at 12:30 UK time, versus 51 of 110 (46%) at other times. Raw records, not adjusted for opponent strength. Only 8 matching games — a small sample.
Man United at 12:30 (away)
History since 2020/21Man United won 6 of 16 away league games kicking off at 12:30 UK time, versus 39 of 102 (38%) at other times. Raw records, not adjusted for opponent strength.
Chelsea v Man United at this ground
History since 2020/21Small sample: 6Since 2020/21: 2 Chelsea wins, 3 draws, 1 Man United wins in 6 league meetings here (2.2 goals per game). Only 6 matching games — a small sample.
Model check: midday kickoffs
Model check, 2024–2026In out-of-sample tests, home teams in this kickoff slot won 48% of 65 games versus 42% predicted. No gap beyond chance: the slot adds nothing the model has not already captured.