Rating form - ideas needed

[PAULM03]

I've had an idea for testing if the form of a team actually has any bearing on the outcome of a game. The problem I'm having though is finding a suitable method for rating a team's poerformance in any particular game. The ideas I've had so far are: League points: This is the one that most people use. A simple league table of the last 6 games to highlight the 'form' teams. The trouble with this is that there is more than one set of results that will give an equivalent rating e.g. 4 draws are equivalent to a win and a draw (i.e. both get 4pts) but this isn't neccesarily the case is it? Goal difference: I'd probably favour this over the league points system. A team with a +6 goal difference from their last 6 games are plainly in better form than a team with a +2GD from their last 6. The problem with this is that it makes no attempt to gauge the number of goals conceded or scored. A team with 6 2-2 draws in a row would have the same rating as a team with 6 0-0 in a row despite one side appearing to have worse defenders and better strikers than the other. Ratio of goal difference to total goals: As an example if a match finished 2-1 the winners would receive 0.66pts whilst the losers would receive 0.33. The advantage of this system is that it takes into acount a teams attacking and defensive capabilities, but again it can't differentiate between high and low-scoring draws and the maximum number of points any team can get is 1, so it won't differentiate between 1-0 and 2-0 etc. So does anyone have any suggestions for a simple metric that could be used to judge a teams performance in a particular game?

[DC]

Re: Rating form - ideas needed Paul, your thread has got me thinking, and while I've been scoffing my tea down I've had an idea. You may not find it useful but here goes... When looking at a teams blind form such as WWDWL and comparing it to DDLLD you'd say the first team deserved more points. But what if you found that the first team had played 4 teams in the bottom 5 and 1 mid-table team, and the second team had played 4 teams in the top 6 and 1 mid-table team. Maybe you could base a teams performance on WHO they played? Take the premiership for example. A team ranked in the top 3 is allocated 5 points, a team based 4th down to 7th 4 points, 8th to 12th 3 points, 13th to 16th 2 points and 17th to 20th only 1 point. This is just an example. Take last night's games: West Ham beat Arsenal - Arsenal were ranked 6th in the table before kick off so West Ham would get 4 points for winning (you could multiply this figure by 1.5 as they were away = 6points) Man City won at home to Newcastle who are ranked in 16th so Man City get 2 points for winning at home. Boro beat Sunderland (away) so score 1 point * 1.5 = 1.5points Does this seem fair? Boro 1.5 points for beating bottom of the table away, Man City beating a slightly better team at home = 2points and West Ham scoring 6 points for winning at Highbury. So basically you get more points for beating a higher positioned team and even more if you win away. You could also lose points for losing to a team in a low position, so if you lost to sunderland at home, reverse the points scale - you'd lose 5 points +1.5 = 7.5 points Also if you lost to Chelsea away, you'd lose 1pt - not too harsh? It makes sense to me but you'd have to look at the teams last 5 point totals perhaps? Haven't considered draws yet. DC