CORDIAL MINUET ENSEMBLE

??????

You are not logged in.

#26 2015-01-06 06:27:56

mzo
Member
Registered: 2014-12-09
Posts: 50

Re: New idea for a tournament structure---feedback, please

Jason, with the exception of the person in 19th, everyone below the top 12 played less than six games and everyone above played at least 6. I know there's probably a variety of reasons, including the one you mentioned. Was just an observation I noticed from when you posted that chart before. I imagine people who won more played more and people who started earlier played more and tended to win more.

Offline

#27 2015-01-06 17:11:58

..
Member
Registered: 2014-11-21
Posts: 259

Re: New idea for a tournament structure---feedback, please

You mentioned Swiss style tournaments. Well, now that I know what a Swiss-style tournament is, it sounds like Elo scoring, limited number of matchs between two accounts, limited number of Elo points transferred between accounts, and most of the other suggested extra rules are all approximations to the bracket system of a Swiss tournament. You can easily drop Elo and have profit-based brackets. Swiss-like tournaments with brackets have a lot of advantages:
* It's a pure profit-driven tournament with a simple extra rule to prevent certain players from playing each other
* Defeats large scale collusion: that would require multiple accounts which also have to play well, otherwise they don't get to a high bracket
* Increases fairness of final ranking: means you can't win the tournament by repeatedly getting paired with and beating a weak player; you have to play other good players
* You can make comebacks. But unlike Elo rating, games at the beginning of the tournament aren't less important than ones at the end
* Unlike preventing players from playing each other more than once, which might result in lots of waiting players, the number of players waiting is limited to the number of brackets, which can be made low, eg 3-5. Instead of rigid brackets you can also make things more flexible, allowing players to play if either their profits are within a threshold (eg within 20 percentile points of the current profit distribution), or if they've been waiting too long (say 3 minutes) without getting a game. Those waits would still foil collusion as immediately quitting games would reset the timer.

Offline

#28 2015-01-06 17:40:56

jasonrohrer
Administrator
Registered: 2014-11-20
Posts: 802

Re: New idea for a tournament structure---feedback, please

Yes, I do like Swiss tournaments a lot.

What I was suggesting was a net-profit-ranked free-for-all where you'd be blocked from re-playing anyone who you'd already won too much money from.

Sounds like you're suggesting something different, where you're blocked from playing people whose net profit is very much different from yours, even if you've never played them before.  Finally, you COULD face someone multiple times after beating them thoroughly once if they climb back up in profit by beating other players (and perhaps you fall down by losing some).

Offline

#29 2015-01-06 17:48:56

..
Member
Registered: 2014-11-21
Posts: 259

Re: New idea for a tournament structure---feedback, please

Yes. Those two systems have advantages and disadvantages to one another, but I don't think any of those are major. So I don't know which to suggest.

A tournament isn't going to "eventually grind itself to a halt" due to exhausting all allowable (unique) pairups if the number of players is large enough, e.g. 40 players, unless we're talking days or weeks.

Last edited by .. (2015-01-06 17:49:36)

Offline

#30 2015-01-06 18:29:36

jasonrohrer
Administrator
Registered: 2014-11-20
Posts: 802

Re: New idea for a tournament structure---feedback, please

Right!

N players is N(N-1)/2 pairings.  20 players is 190 pairings, which, at 5 minutes each, is over 15 hours.

Still, even with an entry fee, some players will drop out by the end if the per-game stakes are non-trivial, which might reduce the pool to a subset who are forbidden from playing each other again.

Maybe this is an argument for making the per-game stakes trivial.  Like, $5 entry fee, but each game is 0.01

I was worried that this would encourage reckless play, but I'm not so sure that it matters.

Offline

#31 2015-01-06 19:44:27

computermouth
Member
Registered: 2014-12-27
Posts: 134

Re: New idea for a tournament structure---feedback, please

If perception of stake worth is the issue: do a $5 buy in, and give players 50 'Rohrerbucks' that can only be used in the tournament big_smile


Try Linux, get free. #!++ (CrunchbangPlusPlus) is a stable distribution based on Debian 8. Keep it fast, keep it pretty.

Offline

#32 2015-01-06 22:18:34

jasonrohrer
Administrator
Registered: 2014-11-20
Posts: 802

Re: New idea for a tournament structure---feedback, please

Only you can call me Daddy Rohrerbucks, computermouth.

But I was trying to avoid having a separate balance that would need to be tracked and displayed.

Offline

#33 2015-01-07 11:52:47

..
Member
Registered: 2014-11-21
Posts: 259

Re: New idea for a tournament structure---feedback, please

N players is N(N-1)/2 pairings.  20 players is 190 pairings, which, at 5 minutes each, is over 15 hours.

No, each of those players can only play 19 games in total; total number of pairings doesn't matter. I was thinking that 40 players is 39 games * 10 min/game = 6 1/2 hours. 10min/game because the rate of games will keep getting slower and slower as you run out of opponents.

But if there's a buyin, doesn't that imply a separate balance? Or would you have you still have to pay stakes for the tournament games from your main account?

I think that effectively ranking players on profit per hour as you suggested is pretty decent. But if you go with a format with no time limit, I suggest setting a limit on the number of games that each player can play. Otherwise if the maximum possible number of games is high then the good (profitable on average) player who plays the most games would have the highest profit, and you could be forced to play for 10 hours to have a shot of winning. Also, you're going to have have some kind of time limit (say, a day or 2) anyway so you can eventually declare a winner.

Last edited by .. (2015-01-07 11:55:08)

Offline

#34 2015-01-07 18:51:52

jasonrohrer
Administrator
Registered: 2014-11-20
Posts: 802

Re: New idea for a tournament structure---feedback, please

Oops, yeah, forgot that those 190 pairings would happen in parallel!

Idea was, for example, a $5 entry fee that is put in the prize pool (taken from your main balance), and then you play $0.50 games from your main balance.  You're blocked from entering the tournament unless you have at least $5.50 (entry fee plus first game stake).

Most profit on $0.50 games over two hours wins.  Prize pool split over various placings.

I'm not sure what the stakes should be.  Lower per-game stakes (like 1-cent) means less risk for players who are falling behind to keep trying.  But also less inherent profit in continued play.  "I have no chance of winning, why would I risk another 50 cents?"  Or "I have no chance of winning, why would I waste my time playing penny games?"

Offline

#35 2015-01-09 00:33:19

joshwithguitar
Member
Registered: 2015-01-07
Posts: 128

Re: New idea for a tournament structure---feedback, please

Is the tournament code currently being tested? I just tried joining a "yellow" 1c game and it told me it would cost 10c to join and would last 4 hours.

Now when I try to join a 1c game I get "unexpected response from server" - is this just the way you join the tournament?

So now do I just start 1c games and wait for others to join?

Will there be prize money on this one?

Edit: Ok, just tried creating a game and got: unexpected response from server.

Last edited by joshwithguitar (2015-01-09 00:35:09)

Offline

#36 2015-01-09 00:41:51

jasonrohrer
Administrator
Registered: 2014-11-20
Posts: 802

Re: New idea for a tournament structure---feedback, please

Fixed!

Offline

Board footer

Powered by FluxBB