1.16 am, Tuesday 16 March 2010

English football league analysis : Detailed results

The English Football Archive provides a superb archive of English football league results dating as far back as the 1888/89 season. The data used here is from the 2002/3 and 2003/4 seasons. Data for the 2004/5 season is included up to 31 March 2005 — which is when the Yardstik analysis was undertaken.

The thick blue line shows how team ratings decrease at a slower rate in the middle three leagues than they do at either top or bottom. The dotted brown line represents* a team's rank.

*Inverted to make comparisons between rank and rating clearer.

This is certainly a "bell-shaped" curve, but with some discontinuities. These may relate to inter-league segregation.
The first half of this table has been derived from league tables published in premierleague.com. The calculation behind the second five columns is decribed in yardstik/football leagues.
The 'yardstik.ini' parameters for each run are listed in the last five column titles.
The two-pass variant of the yardstik algorithm invariably produces a more compressed range of ratings than the single-pass variant. It is for that reason that the latter is favoured in these tables.

A blank value for home advantage does not mean that no home advantage exists, but rather that it has failed the test of significance. It is quite possible that if more evidence were available (i.e. more recent match results were available) then home advantage would be found to be a significant factor in match outcome — in which case a parameter would be printed.

[In any event, the basis of the statistical significance test used (Student's 'T') is fairly dubious and it may well over-estimate significance in all cases.]

No league table were found for conference league clubs. The fixture list doesn't seem to ensure equal numbers of matches for all clubs.

 

 

 

 

 

 

 

 

 
© 2005 Ralph Seeley sitemap Designed by Ralph Seeley