|
|
Post by funktion on Jan 26, 2014 10:09:27 GMT
Can anyone make sense of the way that Challonge calculates breakers? As far as I can tell the tie break score is completely arbitrary, it definitely doesn't use any of the systems which I am familiar with.
Breakers are definitely about to come into play, there are going to definitely be some people that have the same record but don't make the top 8. I'd like to know how this is going to be calculated beforehand, kinda important to have these rules established early on to maintain integrity etc...
Anyways was just hoping we could get some clarification.
|
|
GPrime
HTP Member 
Awesome
Posts: 101 
|
Post by GPrime on Jan 26, 2014 14:40:30 GMT
The system challonge uses is the sum of your opponents' scores, excluding the lowest and highest.
However, John is using different software to calculate the final bracket which uses opponent win % as the first tie breaker. So we won't have completely accurate standings until the last couple rounds when he runs the numbers.
|
|
|
|
Post by funktion on Jan 26, 2014 20:03:14 GMT
Cool, yeah opponent win % seems much better to use.
|
|
|
|
Post by showsni on Jan 26, 2014 20:37:57 GMT
And I guess ties count for nothing? As in, on challonge, someone with 3 wins, 1 loss and 1 tie can be below someone with 3 wins and 2 losses because of tiebreakers.
|
|
|
|
Post by Werlix on Jan 26, 2014 21:35:18 GMT
Cool, yeah opponent win % seems much better to use. Opponent's win % is the same as what is being used now (opponent's score which is directly related to win%). The only difference is that it excludes the best and worst results. Seems weird to circumvent the challonge leaderboard and redo custom tiebreakers just to reintroduce the best and worst results. |
|
GPrime
HTP Member
Awesome
Posts: 101
|
Post by GPrime on Jan 26, 2014 21:38:36 GMT
And I guess ties count for nothing? As in, on challonge, someone with 3 wins, 1 loss and 1 tie can be below someone with 3 wins and 2 losses because of tiebreakers. A tie is the same as a loss because the only "ties" in this tournament are no-play double-loss. If someone didn't play a round, there's no reason for them to be ahead of someone who played but lost but otherwise has the same number of wins. Both a tie and a loss are the same number of points (i.e. zero.) Opponent's win % is the same as what is being used now (opponent's score which is directly related to win%). The only difference is that it excludes the best and worst results. Seems weird to circumvent the challonge leaderboard and redo custom tiebreakers just to reintroduce the best and worst results. Now quite, werlix. For example, if someone played two games (1 win, 1 loss) then dropped the tournament, they'd be worth 1 point in the current tiebreaker system. If you go by win %, they're worth 0.5 which makes them worth more overall. The win % system avoids penalizing people for their opponents dropping out early--something that's certainly not your fault. |
|
|
|
Post by Admin on Jan 27, 2014 1:39:48 GMT
The system that challonge uses is the same system that chess uses for tie breakers. It's called the Median-Buchholz system. What it is is the sum of your opponents wins and losses minus the best and worst results. I'm going to plug the results into Mantis and use opponents match win % as the first tie breaker to sort out the mess of 6-2's.
- John
|
|
|
|
Post by htpFlyingMeatchip on Jan 30, 2014 3:50:22 GMT
When can we expect results from the Mantis software?
|
|
|
|
Post by Admin on Jan 30, 2014 4:03:09 GMT
When can we expect results from the Mantis software? I posted the tie breakers for the top 25 players in the "contact your opponent" forum as a sticky. Right now, Mantis is being a bastard and won't let me export the full standings from the program so once I figure that bug out I can post them in full. For now, I hand typed the top 25. - John |
|
