# EI Match Probabilities for the English Premier League

Posted by Martin Eastwood May 17, 2013 8 Comments 1823 views

We have finally reached the end of the season so for the last time in 2012-2013 here are the Eastwood Index’s (EI) probabilities for the English Premier League.

Once the season is over and done with I’ll be looking back at how the EI has performed and how well it’s predictions compare with the bookmakers so look out for that next week!

 Home Team Away Team Home (%) Draw (%) Away (%) Chelsea Everton 52 28 20 Liverpool QPR 71 18 11 Man City Norwich 77 14 9 Newcastle Arsenal 23 32 45 Southampton Stoke 42 31 27 Swansea Fulham 46 30 24 Tottenham Sunderland 67 21 13 West Brom Man United 13 30 57 West Ham Reading 49 29 23 Wigan Aston Villa 43 31 27

Martin is football fan and data scientist. In his spare time he likes to combine the two and write about the mathematical analysis of football.

1. - May 17, 2013

Cool website. I really enjoy making projections and football seems to be the more interesting sport to do this because of the variety of approaches (given the low scoring nature of it which makes it less predictable). I go the least squares way and assume error is normally distributed around predictions from a rating system (and its the only one I could do easily in Excel given my limited maths). I think the guys at DTech go the ordered probit/logistical regression route (which seems to be the way to go I think) and I haven’t figured out your method yet but it looks interesting. They all seem to have similar numbers though:

e.g.: for Chelsea/Everton (Home/Draw/Away)

DTech: 55/23/22
Least Squares: 59/22/18

All of which are around the bookmaker odds for that game 60/24/19 (random bookmaker picked)

Out of interest, have you noticed any kind of preferred ratio of your projections to bookmakers odds along the lines of the Dixon/Coles paper into this kind of thing?

Keep up the good work.

• - May 18, 2013

Thanks George! Once the season finishes I’m planning looking back at how I’ve done over the year compared with the bookmakers to see if there are any patterns to my projections versus theirs.

• - May 18, 2013

Thanks. Re: the bookmakers that’s what I’ve found, just because you can generate a number similar to theirs – what does it actually tell you (as we don’t know what their perspective is)? What biases is it accounting for? Don’t know if you saw the Steven Levitt paper in 2004 on this kind of thing (was on the NFL though), and the various papers done on football (e.g. Graham and Stott from DTech in 2008) and bookmaker efficiency. I know DTech have worked out they would make something like 10% when their number differed from bookmaker numbers over however many years they have been doing it. The Dixon/Coles paper also found that when the ratio of their probability against the bookmakers probability was about 1.2 it was optimum for generating profit. What this tells us though – I don’t know.

Good luck with everything and I look forward to reading it.

• - May 21, 2013

Thanks George, I’ve not seen the Levitt paper before.

2. - August 23, 2013

Very interesting, been trying myself to see how accurate the bookies have been with their odds. Scraping my data off Oddsportal, using a Java app to analyse it. And then generatinig my own rating based on form etc
Only point of note I have noticed so far is that home teams with odds between 1.999 and 2.5 are the best to bet on. Consistently generating a % profit across all leagues. I assume these are games where bookies are loosest with their odds.

• - August 23, 2013

yes the bookies are pretty good but they are certainly not perfect and there are situations where they can be beaten – the tricky bit is being able to do it consistently over a long period of time

3. - August 23, 2013

I’d like to see you try your system against the Iranian pro league. Very obscure league to be betting on I know but I have found my own rating system giving me accuracy of 70% with average odds of over 2.2.
I was raking it in last season. Not so much this time round as the bookies seem to have tightened their odds on that particular pony.

• - August 23, 2013

cool, well done it’s always great to hear about people taking money off the bookies

I haven’t tried my system outside of Europe or MLS yet, now you have got me really interested in trying it out on more leagues!

• #### Massey Ratings For Football Part Two

December 4, 2014 / 6 Comments
• #### Massey Ratings For Football Part One

November 27, 2014 / No Comments
• #### English Premier League Pythagorean

November 4, 2014 / No Comments
• #### Predicting Football Using R

November 2, 2014 / 11 Comments
• #### Expected Goals: Foot Shots Versus Headers

August 28, 2014 / 10 Comments
• #### Expected Goals For All

February 12, 2014 / 23 Comments
• #### Applying Elo Ratings To Football

February 7, 2013 / 20 Comments
• #### Rating Teams and Predicting Football Matches Using the EI Index

February 21, 2013 / 20 Comments
• #### How Accurate Are The EI Football Predictions?

March 21, 2013 / 20 Comments
• #### Understanding Total Shot Ratio in Football

April 2, 2013 / 20 Comments
• #### Martin Eastwood

Good idea, will post a follow up when I get chance …

• #### marko

teams of 12 players with requirement that one play …

• #### Martin Eastwood

Cool, look forward to reading it :-) …

• #### Peter

Hi Martin, I have given it a go (through Excel, …

• #### Martin Eastwood

It was all taken from my PostgreSQL database so yo …