- 20. Norwich City
- Points: 21
- Predicted points: 28.3
- 19. Aston Villa
- Points: 25
- Predicted points: 32.6
- 18. Bournemouth
- Points: 27
- Predicted points: 34.3
- 17. Watford
- Points: 27
- Predicted points: 35.4
- 16. West Ham
- Points: 27
- Predicted points: 36.4
- 15. Brighton
- Points: 29
- Predicted points: 36.5
- 14. Southampton
- Points: 34
- Predicted points: 45.1
- 13. Newcastle United
- Points: 35
- Predicted points: 46.3
- 12. Everton
- Points: 37
- Predicted points: 48.4
- 11. Crystal Palace
- Points: 39
- Predicted points: 49.6
- 10. Burnley
- Points: 39
- Predicted points: 51.1
- 9. Arsenal
- Points: 40
- Predicted points: 53
- 8. Tottenham
- Points: 41
- Predicted points: 54.2
- 7. Sheffield United
- Points: 43
- Predicted points: 57.3
- 6. Wolves
- Points: 43
- Predicted points: 57.4
- 5. Manchester United
- Points: 45
- Predicted points: 60.4
- 4. Chelsea
- Points: 48
- Predicted points: 62.9
- 3. Leicester City
- Points: 53
- Predicted points: 70.8
- 2. Manchester City
- Points: 57
- Predicted points: 77.8
- 1. Liverpool
- Points: 82
- Predicted points: 105.9
“For a given match H-A, with H the home team and A the away team, let us model the number of points nH won by H as a function of the current points difference pH - pA. The larger the points difference (for instance, pH - pA = 82 - 25 = 57 for Liverpool - Aston Villa on match-day 33), the larger nH and the closer nH to 3; nH increases from 0 to 3," claims the report in The Times.
“Similarly, the number of points nA earned by A is described as another function of pH - pA, which decreases from 3 to 0. I suggest using functions known as sigmoids or logistic functions. Each of these two functions depend on two parameters (elasticity and home advantage) which are calibrated to the 2019-2020 Premier League results.”
Source: GMS