2005 Summer League

Send Results and Corrections to Si-lam (sichoy@excite.com).

August 25, 2005

For entertainment purposes only, see computer rankings.

	Num	Team	Team Name	Win	Loss
Competitive Division	1	5	Booty and the Ghost Dogs	10	0
	2	2	Palpatine's Posse	8	1
	3	12	Tenacious D	7	3
	4	9	Lemon Jello	7	2
	5	1	Big Discs	7	4
	6	6	Righteous Harmony Fist	5	5
	7	10	Your Mom	5	6
	8	11	FUNdamentals	3	7
	9	3	Ask Us	2	7
	10	7	Fish Heads	2	7
	11	4	Penguin Army	2	8
	12	8	Talon	1	9
Spirited Division	21	24	Homegrown	11	0
	22	31	Ultimately Irresponsible	9	2
	23	32	Westwinded	8	3
	24	21	Alchemists	6	4
	25	27	New Dogs Old Tricks	5	5
	26	25	Intergalactic Alliance	5	5
	27	30	Turbo Slugs	4	7
	28	28	Stuffed Animals	4	7
	29	23	Flying Chipmunks	4	7
	30	26	Monkey Mayhem Crew	3	7
	31	29	Toby's All Stars	2	8
	32	22	Defenestrators	1	7

Rankings are based on published results and only apply within a division. They only represent the strength of teams based solely on past performance, and no-show games are excluded. The ratings are relatively poor predictors because the presence/absence of key players can significantly alter the outcome.

The methodology is: Every team has a strength variable. For each outcome of a game, a simple and rather arbitrary formula is used to quantify the probability of the winning team winning again if another game is played with the exactly the same conditions and personnel. The probability formula is (winning score + 5) / (winning score + losing score + 5). This value is then compared with a predicted probability based on the difference in strength for both teams. The predicted probability is defined as the standard normal cumulative distribution function of (winner strength - loser strength). Now the difference between the game result probability and the predicted probabilty is the error. A least sum square fit of all the errors for each game is performed with each team's strength as an independent variable. The outcome are the strength ratings for each team that minimize the errors between predicted and actual performance. Finally, the strength ratings are scaled to about 0-15 points.