As the simplest metric of forecast performance, the table below gives the number of correct scorelines and results according to the model’s most likely scoreline predictions through the 2018/19 season.
“Lawro” points follow the scoring system from BBC’s Mark Lawrenson weekly predictions: 40 points for a correct scoreline; 10 points for a correct result.
We also compare the model’s cumulative performance against Mark Lawrenson throughout the season.
| Scorelines | Results | “Lawro pts” | ||
| Model | Lawro | |||
| Total | 38/380 | 210/380 | 3240 | 3340 |
| 38: 12 May | 1/10 | 5/10 | 80 | 70 |
| 37: 3-6 May | 1/10 | 7/10 | 100 | 90 |
| 36: 26-28 April | 1/10 | 3/10 | 60 | 60 |
| 35: 19-21 April | 1/10 | 4/10 | 70 | 60 |
| 34: 12-15 April | 0/10 | 6/10 | 60 | 130 |
| 33: 4 April | 1/10 | 6/10 | 90 | 100 |
| 32: 30 Mar-1 Apr | 1/10 | 7/10 | 100 | 170 |
| 31: 16-17 Mar | 0/10 | 4/10 | 40 | 80 |
| 30: 9-10 Mar | 0/10 | 6/10 | 60 | 30 |
| 29: 2-3 Mar | 1/10 | 7/10 | 100 | 100 |
| 28: 26-27 Feb | 0/10 | 8/10 | 80 | 60 |
| 27: 6-24 Feb | 2/10 | 6/10 | 120 | 160 |
| 26: 9-11 Feb | 2/10 | 6/10 | 120 | 140 |
| Rounds 1/25 | 27/250 | 135/250 | 2160 | 2070 |
[Note: Up to 17th August this evaluation was based on the most likely (unconditional) score forecast. From 17th August, as we refine our forecasts, it is based on the most likely result and subsequent (conditional) score forecast]