Free Allegiance Forums

http://asgs.alleg.net/asgsnet/recentgames.aspx?gameid=14907

This is what gets me about ELO. I was just browsing tonight's games and came across this gem. I'm sorry but ELO will never work as long as games like this are counted.

The threshold, in my opinion, is set too high. A 70/30 game is already very difficult to win.

Oh, I think 80/20 is more than good. /smile.gif" style="vertical-align:middle" emoid=":)" border="0" alt="smile.gif" />

jgbaxter wrote:QUOTE (jgbaxter @ Sep 7 2006, 06:51 AM) Oh, I think 80/20 is more than good. /smile.gif" style="vertical-align:middle" emoid=":)" border="0" alt="smile.gif" />

Do I read this correctly, that you would consider 4:1 winchances as ok? ;-)
(Though this is certainly a subjective call to make)
I for one would consider the threshold of fairness exeeded at above 60/40 (1.5:1)

60/40 games might be still fun to play, but where is only honor in beating the enemy if chances to win were not equal? (and I do not refer to the type of equal with huge tolerance margins ;-)

Perhaps you forget elo awards are variable?

I think a 79.9 v 20.1 game is worth 2 elo. 10 wins like that and you can afford to lose once... of course most people don't have a 89% winning percentage... so stacking at even that degree doesn't pay.

jgbaxter wrote:QUOTE (jgbaxter @ Sep 7 2006, 04:26 PM) Perhaps you forget elo awards are variable?

I think a 79.9 v 20.1 game is worth 2 elo. 10 wins like that and you can afford to lose once... of course most people don't have a 89% winning percentage... so stacking at even that degree doesn't pay.

It's worth (1.0-(1*.799)) * 32 = 6.4 Elo, actually.

And my point is that while the system is designed such that it would be a 4:1 win ratio, in practice, it's not. When Elo is comparing two individual players, all it can take is one screwup on the high-ranked person to blow it for them, whereas a team of high-ranking players is less likely to all screw up in the same game.

That game was pathetic.

~4

just squeaked that one by, i had to wait 13 seconds before allowing shiz to join, to make sure the ELO was ok.

Terra, it was just a number I tossed out, but oh well, exactitude is fun too.

Until something is done where one player by themselves can't stack a 30 player team, I suppose it's kind of funky anyway

My suggestion/opinion would be let the system run for 60 days. Let's see where the ranks, etc balance out to. Then if it still appears to be an issue, we should address it then.