Olympiakos
Model pick
46%
Fair odds 2.18
Milan
26%
Fair odds 3.84
Share of the pitch
Who wins
| Olympiakos | Draw | Milan | |
|---|---|---|---|
| Goals model | 46% | 28% | 26% |
| Elo model | 42% | 28% | 30% |
Why this prediction
-
Overall strength Favours Milan
Milan are rated 21 points above Olympiakos (1575 to 1554), built from every result, weighted towards recent ones.
-
Home advantage Favours Olympiakos
Playing at home is worth about 65 rating points, which adds 9 points to Olympiakos's chance of winning and takes the same from Milan's.
-
Attack against defence Favours Olympiakos
Expected goals 1.4 to 1.0. Olympiakos score 1.2 a match against an average side (3rd in Greek Super League); Milan concede 1.2 (8th best defence in Serie A).
-
Last 5 matches Favours Olympiakos
Olympiakos DWWWW, gained 22 rating points; Milan LWLWL, lost 22 rating points. Already counted in the ratings above.
Rating, last 12 months
OlympiakosMilanOlympiakos
3/10 over 2.5
16 scored
4 conceded
09/12
09/16
09/20
10/11
10/15
10/18
10/22
10/25
10/31
11/04
Milan
6/10 over 2.5
18 scored
15 conceded
09/16
09/20
10/11
10/15
10/18
10/22
10/25
10/28
10/31
11/05
Bars show total goals in each of the last 10 matches, oldest on the left: the team's goals in its colour, conceded in grey. The dashed line is 2.5 goals.
Total goals
Both teams to score & clean sheets
Most likely scores
- 1-1
- 1-0
- 0-0
- 2-1
- 2-0
- 0-1
- 1-2
- 2-2
Every score
| 0 | 1 | 2 | 3 | 4 | 5 | |
|---|---|---|---|---|---|---|
| 0 | 9 | 8 | 4 | 2 | ||
| 1 | 12 | 13 | 6 | 2 | 1 | |
| 2 | 9 | 9 | 5 | 2 | ||
| 3 | 4 | 4 | 2 | 1 | ||
| 4 | 2 | 2 | 1 | |||
| 5 |
Half-time result
First-half goals
Half-time / full-time
| Half-time ↓ Full-time → | Olympiakos | Draw | Milan |
|---|---|---|---|
| Olympiakos | 25%4.04 | 3%30.03 | 1%— |
| Level | 19%5.39 | 17%5.81 | 8%12.55 |
| Milan | 3%31.25 | 6%15.72 | 18%5.61 |
Corners
Most corners
Corner totals are hard to call: over 9.5 the model's log loss is 0.685 against 0.693 for a naive guess, across 30,937 past matches. Which side wins more corners is called much better.
Cards
Over 4.5 cards, the model's log loss is 0.653 against 0.682 for a naive guess, across 33,146 past matches. The referee matters a lot and isn't in the model.
They haven't met in the data.