Correct scores (what? again?)

[Data]

Here's a bit of speculative fun, with looking at making a profit from one of the most difficult of football markets.

Using Excel's "Solver" I processed the Premier league results to date grid by allocating an adjustable variable for each team's home/away attack/defence ratings to drill down to the best fit of correct scores to the games played to date. Once the best fit is arrived at then simply applying the finished formula to the next set of games.

I figured (rightly or wrongly) that the strategy would need at least half of the season completed in order to have a healthy sample size to be anywhere near "reliable", for this reason I'm only concerned with games from January 1st onwards. The two game days so far this season have given 3 correct from the 20 games played, bet365 odds for these were 8.00 (twice) and 9.00 from indicating two 1~1 results and a 2~1.

Thus far then 25pts return from 20pts staked. Here's this weekend's Excel predictions, can it keep up the profitable momentum?

Watford		Tottenham		2~2
Arsenal		Sheff United		0~1
Brighton		Aston Villa		2~2
Man City		Crystal Palace		1~1
Norwich		Bournemouth		2~1
Southampt		Wolves		1~2
West Ham		Everton		2~1
Newcastle		Chelsea		1~2
Burnley		Leicester		0~2
Liverpool		Man United		2~0

[Zico10]

I love a correct score prediction. I'll follow this with interest! Good luck