Here are the latest match probabilities for the English Premier League calculated using the Eastwood Index (EI).
Somewhat surprisingly, Liverpool are only just favorites to beat Fulham with the odds so close that a draw would seem the likely outcome.
Down at the bottom of the table Newcastle versus QPR and Norwich versus West Brom look likely to finish tied, while Sunderland are slight favorites against Southampton meaning Wigan desperately need to take points off Arsenal to stand any chance of avoiding relegation.
| Home Team | Away Team | Home (%) | Draw (%) | Away (%) |
| Aston Villa | Chelsea | 21 | 32 | 47 |
| Stoke | Tottenham | 24 | 32 | 44 |
| Everton | West Ham | 67 | 20 | 12 |
| Fulham | Liverpool | 31 | 33 | 36 |
| Norwich | West Brom | 37 | 32 | 31 |
| QPR | Newcastle | 32 | 32 | 35 |
| Sunderland | Southampton | 49 | 29 | 22 |
| Man United | Swansea | 76 | 15 | 9 |
| Arsenal | Wigan | 71 | 18 | 11 |
| Reading | Man City | 8 | 28 | 64 |
Acka Bilk - May 10, 2013
Interesting to see that you make Sunderland such strong favourites against Southampton. On what basis can that be? Are you just looking purely at very recent results and the league table?
Martin Eastwood - May 10, 2013
It is based on results over the past few seasons, with greater weighting applied the more recent the result is. The two team’s are rated reasonably similar at the moment so part of Sunderland’s advantage is likely due to playing at home.
Saze - May 14, 2013
If the elo ratings are based on results over the past few seasons then how do they apply to newly promoted teams that have never played in the premier league before? Also when you say “over the past few seasons” how many seasons are you exactly talking about.
And have you ever considered crafting elo ratings for individual players and then using this to create an elo rating for a whole team as team lineups do usually vary from match to match and can sometimes significantly affect the outcome of a match. The downside to this however is that team lineups are only announced 30-45 minutes before the match starts so you won’t really be able to make a prediction until the match has practically started although it can still be useful to look at which players are performing at the very top level.
Martin Eastwood - May 14, 2013
My EI ratings are currently based on three previous season’s data plus this season. Over the summer I hope to look at whether extending this offers any advantages.
Newly promoted teams are tricky. For the EPL teams are assigned the relevant relegated team’s rating so the team promoted in first place get the team relegated third from bottom’s rating. It’s not perfect but over time the rating will correct itself and move towards the correct value. For the MLS there is no relegation to worry about so I can avoid this.
I am interested in trying to model teams based on their players but this still requires more research into what stats to use per player – passes, tackles, shots, etc etc??
Saze - May 14, 2013
I understand that a promoted teams elo rating will correct itself over time but wouldn’t just creating elo ratings for the Championship, league one and league 2 overcome this issue.
Moreover if you are interested in individual player elo ratings then castrol rankings is a good player ranking site; “http://www.castrolfootball.com/rankings/rankings/?team=&comp=&nation=&position=&search=&offset=0&jump=1 ” You can find out more on how they create the rankings in their FAQ section.
Martin Eastwood - May 14, 2013
I’m not sure it would really work. For example Cardiff’s EI would be based on how they have performed against Championship quality teams so would not reflect how they would perform against Premier League teams.
Goalimpact - May 15, 2013
‘Individual player ELOs’
This is what I do. It works quite well, however it is far from being easily implemented. If you are interested please drop by my blog Goalimpact.com
Saze - May 16, 2013
Very interesting. I like the “top-down” approach that you have taken.
Submit your comments below, and feel free to format them using MarkDown if you want. Comments typically take upto 24 hours to appear on the site and be answered so please be patient.
Thanks!