Tuesday, 8 June 2010

Elo Overall


The ratings were applied across a total of 3,763 matches. Only the later season matches in Europe (a total of 34) were included, so while these are included in the overall totals, they are excluded as an individual league on account of the low sample size.

Across all leagues, the overall strike rate, i.e. the final margin was exactly as predicted, was 25.91%, with the leagues ranging from a high of 29.4% in Ligue 1 to a low of 21.27% in the Conference National.

For matches where the predicted result was correct, but not necessarily the margin, the overall win rate was 45.97%, ranging from a high of 53.16% in the Premier League to a low of 36.94% in the Scottish Second Division.

Looking at draw selections, they hit overall at 27.87%, with the best league being the Premier League at 38.67% and the worst being the Scottish Third Division at just 21.63%.

When the ratings have an expected winner, the more cautious might choose to lay the opposed team. The best leagues for this strategy are La Liga and Serie A with just 9.56% and 9.79% respectively of selections actually losing. The worst? Scottish Third Division at 19.3%. Of the five top leagues, the Bundesliga is the poorest at 16.67% with no less than 29% of expected one goal winners losing. A finding that echoes Talkbet's observations.

Already, there's a trend developing suggesting the lower leagues should be dropped, but that's looking at the numbers from a backing perspective. There's more than one way to look at these numbers.

3 comments:

  1. Your draw hit rate seems purely in line with expected % of draws. Fluctuations from League to league are obviously to be expected in the relative short-term and short-medium term.

    Whatever your short-medium term results I don't see how you can expect such a model to be profitable in the long-term if you're not actually pricing up matches yourself and are simply going on the way you use your ELO model (seemingly with scant regard for team news etc.)

    ReplyDelete
  2. Have you thought of fleshing out you data but using the historic odds on say football-data.co.uk instead of the fairly meaningless percentages?

    ReplyDelete
  3. How do these strike rates compare to average prices you did(/could) have obtained?

    ReplyDelete