Bughouse Live Tournaments - a possible design

Bughouse live tournaments are more complicated than normal live tournaments, due to the fact that it is a team game.  You need to be able to handle drop-outs, odd numbers of participants, etc.

There are two main approaches: team tournaments (where you keep your partner all the way through), and individual tournaments (where you get a different partner every game).

Both approaches have their merits, but I suspect the the team tournament, where you have the option to choose a partner or get one assigned randomly but keep your partner for the entire tournament, would be more popular.  However, I believe it has a few more complications than the individual tournament, so maybe the initial implementation should go with the individual tournament, and team tournaments can be added later.

(One problem with team tournaments: if your partner bails out after a couple games, you're stuck without a partner.  Are you then forced out too?  You could then be paired up with someone else who lost their partner, but how do you report that in the standings?  Whose results from the first two rounds does the new team get - the better set, or the average of the two?  How do you do tiebreaks for someone who beat one of the two original teams?  I think a solution can be found for all the problems, but it will take some work.)

So, assuming an initial implementation of individual tournaments, here's a possible design.

1. You get a new partner randomly assigned on every round (although not fully randomly, as described below).
2. In a normal swiss-style tourney, you are matched against someone with a similar score.  For bughouse, you should be paired with someone with a similar score.  Thus, a strong player may be saddled with a weak partner in the early rounds, but this will be rare in later rounds.  The top games in later rounds should generally have four strong players.
3. In a normal Swiss-style tournament, you cannot meet the same player twice.  However, for bughouse, you should be able to end up in the same game as another player a maximum of three times: once as partners, once as opponents on the same board, and once as opponents on the opposite board.
4. Once a tournament has started, new players should be allowed to join as reserve players.  If, due to dropouts or an odd number of initial entrants, there are stranded players with no partner or opponent, players will be drawn from the reserve pool to make up the numbers.  The system should, where possible, try to choose players that, based on their rating, will result in a roughly even match.  These reserve players will remain for the rest of the tournament unless they too drop out; thus, a reserve player who joins early enough and then wins most of their games could theoretically win the tournament.  If there are not sufficient reserve players to complete a match, they should remain in the reserve pool and any stranded players should get a bye.
5. Tiebreak should be based not only on the scores of your defeated opponents but also on the scores of your winning partners.  A good way to do this would be to add the scores of all your defeated opponents and subtract the scores of your winning partners.  If Adam and Ben beat Charles and David, Adam's tiebreak from that game would be the number of games Charles won, plus the number of games David won, minus the number of games Ben won.  This means you are rewarded both for beating strong opposition and for winning when partnered with a weak player.  (Note: this means negative tiebreaks are possible if a weak player's only win comes when partnered with a strong player against two other weak players.)

Any complications I've overlooked?  Any other thoughts/comments?

A lot of thought must have gone into that!

The number of players must be a multiple of 4 to allow everyone to get a game; many tournaments I play in only get around 10 players, so there may have to be as many as 3 reserve players able to come in at any point, or up to 3 people may end up getting a bye round. Say you have 11 players and no reserves, every round 3 people will get a bye; by round 4 someone will have to take a second bye. Imagine signing up for a 5 round tournament and getting only 3 games through no fault of your own.

Also, would colour rotation affect the pairing of partners/opponents?

Regarding your tiebreak, there might be a few flaws. Say your partner is strong, and both opponents are also weak. If your partner's opponent were to blunder before your opponent, or your partner's opponent times out, you would be penalised in tie-break heavily. Why should you be penalised for being unlucky?

Maybe there could also be a provision for matches that were lost on your board, so if you made the blunder that lead to your team's loss, that affects your tie-break? This could run into the problem that if your partner massively blundered (e.g. hung a queen) that got passed onto your opponent, who immediately drops it into mate, who should be penalised, and how would that be judged? This could even be taken advantage of, if in the last round you are paired with someone you are jointly leading with (tie-breaks also equal), you could blunder deliberately, have your opponent take the 'on board loss' and yet you could win overall!

One idea for team tournaments, maybe you should only be able to enter with someone you're friends with, so that there is a greater trust between you, and so there would be a lesser chance of a withdrawal.

I'd join, 

I can see advantages and disadvantages of team tournaments compared to individual tournaments, but in my mind; if team tournaments would end up more popular, then doing individual tournaments as a first implementation is a bad idea. Temporal implementations can easily turn out permanent and some things you have to figure out in individual tournaments such as the pairings and tie breaks can be seen as a waste of time if your aiming for making team tournaments later anyway. If there is a change from individual tournaments to team tournaments later it would also cause a bit of confusion among the players. Some might argue having both implementations is an option, but that would reduce the number of players that would join each tournament since people will prefer one over the other. I believe doing it right from the start is the way to go.

Overall I think your thoughts about individual tournaments are quite good and clearly a lot of thought behind it, but as I said earlier; I think team tournaments are the way to go.

I definitely think people would prefer team tournaments. The individual tournaments also sound good but if it's a choice between the two, team tournaments are probably preferable. 

@ThomasJEvans:

For tournaments to work at all, they need to attract enough interest.  You probably need about 12 players minimum for a tournament to work.  But a championship where top players are invited would attract a lot of people; the crazyhouse qualifiers for the upcoming tournaments have been attracting dozens of people - I've seen 80ish in one of them.

For colour rotation, not sure why that would affect anything.  I'm not sure colour rotation needs to be anything other than random, given that both sides have both a white and a black.

On tiebreak, with my design you're penalised for winning with a strong partner against weak opposition.  It doesn't matter whether you win because there's a flag, a blunder, whatever, you're winning a game you're expected to win.  It's just like beating a minnow in a normal tournament: you don't get much tiebreak out of it, and it doesn't matter if you win because your opponent hangs their queen or flags.  It seems intuitive to me that your tiebreak should be worse than someone who beats similar-quality opposition with a weak partner.

It may seem like a penalty, though, because we're subtracting something.  But remember that we're subtracting something for everyone, in every single round, and that tiebreaks are all relative.  So it's not really that it's a penalty, it's that there's an assumed subtraction that you can avoid if you manage to win with a weak partner.

I also don't think it's a great idea to try to determine if it was lost on one board.  For one thing, there's a huge grey area in which it's not clear on which board a game was lost.  Perhaps I blundered my queen because I was under pressure because my partner kept hanging knights.  Perhaps I got flagged because I was sitting because you needed me to stop losing pieces for 30 seconds.  I think we just need to assume that both boards live and die together.

I also don't see much reason to worry about deliberate losses.  It would be extremely rare, I think, for a deliberate loss to steal the tournament win from your partner - remember that a loss will cost you a whole point.  And even in one of those rare circumstances, chances are very high that neither player will do the calculations needed to determine that it is possible.

Here is one contrived scenario where it could happen:
Final round of the tourney.  Adam and Bob are two points clear of the chasing pack, tied on points and tiebreak.  They are partnered in the final round.  All other games in the final round are completed, so they can see (with 0:30 on the clock) that they are still exactly tied, and the only thing that can change that is tiebreak from this game.  They are playing against Charles and David.  They have both beaten both Charles and David once previously, so the only difference between them is that Adam has partnered successfully with both Charles and David, while Bob has only partnered successfully with Charles (maybe he's not partnered with David, or maybe they lost; doesn't matter).

Bob decides to intentionally throw the game, giving Charles and David a point.  This gains +2 tiebreak for both Adam and Bob for having beaten both Charles and David once previously.  It costs Adam 2 points for having partnered successfully with both Charles and David.  It costs Bob 1 point for having partnered successfully with Charles.  Thus Adam nets even, and Bob gains a net +1 to his tiebreak, for their wins against and in partnership with Charles and David.

You see how much of a stretch that scenario is?  You have to be more than one full point clear of the chasing pack, you have to remember who you've partnered with and played against (including opponents on the opposite board), and you have to know who your partner has partnered with and played against, and you have to calculate it all out - all with the clock ticking.  I don't really see much reason to be particularly concerned here.

On team tournaments, I focused on individual tournaments initially to avoid that problem.  But I don't think only allowing partnering with friends is a great solution, either.  I think you either need to assume that if one partner bails out that both are gone, or else you need to try to find a way to replace the missing partner (reserve player would be one option, matching widowed partners is another option if the complications can be resolved).

@Martin0 and @helmsknight, I do think team tournaments would be more popular, but I think there's merit to both.  I don't really see any reason why both can't be implemented.  If they do not run at the same time (or particularly close to each other), there's no reason not to have both.

@MGleason Yes, agreed. Both would be great to have. 

Good ideas. If you have a multiple of 4 people, another tournament style that has been run in the past is something we called a "Grand Prix."

The way it works is that in every group of 4, you play all 12 possible pairings - one game with each partner against each opponent with each color.

If you have 16 players, that's 4 groups playing their own round robin. Then, depending on how long you want the tournament to go, you can drop the bottom N players and start the next round with either 4, 8, or 12. Tiebreaks are often necessary.

When you're down to the final four, they play their 12 games, and then the tournament is over.

This format takes a while - 12 games minimum. It may be appropriate for a more serious event (weekly scheduled tournament, individual bughouse championship, etc.) rather than just a random pickup tournament.

imagine correspondence bughouse LOL

@happytoad, that sounds interesting!  But it probably works best with several qualifying rounds across multiple days if done online.  It probably wouldn't work for a random pickup tourney.

@AntonioEsfandiari, there's no reason the other variants wouldn't work as correspondence.  Correspondence crazyhouse would be fun.  In bughouse, however, the clocks on the two different boards are so critical, and sitting for a few seconds is sometimes part of the strategy.

That said, I have heard of people playing correspondence where the move alternates not just white/black as in a normal game, but also from one board to the other.  Thus, white moves on board 1, then white moves on board 2, then black moves on board 1, then black moves on board 2...

However, without the clocks, it would be a very different game.  Sitting for a few seconds wouldn't be an option.  It would also be much easier for two partners to communicate via message and analyse a position much deeper - thus, in effect, resulting in the stronger player driving both games and picking virtually every move.  That would kind of ruin part of the purpose of bughouse, as the teamwork element would be gone, and the weaker player would be left with little or nothing to do other than physically make the moves the stronger player fed them.

I used to be a Tourney Manager with a bot called gparbiter in fics. It worked in the same way that grand prix and was really interesting for everyone who has played. I recommend this format and first 2 of each group going to new groups 16-8-4-1

these are like whole paragraph comments stop it peeple

lol jk

You people are imagining possible setups for bughouse tournaments as if we haven't been running and participating in actual bughouse tournaments for the last 20 years.  It's ridiculous!!  chess.com gets bughouse and you all act like bughouse was just invented.

The GPArtbiter(TD) bot on FICS worked very well for individual bughouse.  Pairings were great.  The bot forms groups of 4 players.  You play 4 games (or 8) with one partner, then switch partners until you've played with every partner.  Top 2 from each group advance.  Form groups of 4.  Repeat.  In theory if there are an uneven number (not multiple of 4), some people could get byes from the early rounds automatically advancing with an average score.

The main problem was it wasn't automatic... it would tell players who to match and partner, but the players would sometimes screw it up.  This all would have been solved by implementation of an rpartner tourney bot command on FICS like the rmatch command.  Unfortunately the admins and programmers on FICS don't give a damn about bughouse, so it never happened after 10 years of asking.  Individuals dropping out could be annoying, but you could always substitute a reserve player.  That would skew the results a little.  The long term solution to drop outs is not fancy programming, but rather a penalty being paid, with abusers labeled and repeat offenders banned from future tournaments.  The same pairing system could easily be used OTB.

Team tournaments work under all the standard tournament pairing schemes.  In Berlin we use a mix: a 9 round swiss qualifier with 4 games per round (switch color and opponent), 8 top teams progress to a single elimination ladder best of 13 games, teams that didn't qualify play a round robin B final with 4 games per round (switch color and opponent).

Here is one example (change the year for many others):

http://chess.emrald.net/bughouse2015/

Stop reinventing the wheel and look around at what already exists.

Thank you, @chuckmoulton. Thank you. 

@chuckmoulton, that's a great design if you have a manageable number of people.  But to run a big tournament like they're currently doing for crazyhouse, to get from ~100 players to 1-2 winners, it will take too many rounds.  With a huge number of people, you're either running a very lengthy tournament (which could in be done with multiple qualification rounds over multiple days), or else you need to figure out a way to make bughouse work in a Swiss-style tournament.

The tournament should be timed by time, not game number. Matches would be made as soon as all the games finish (This style could work for both individual and team tournaments). The timed would be better so if the time is one hour you might get 6 games in or 10. That would be better so people could know how to plan for the tournaments, instead of having to drop outing the middle of the tournament.

@aa-ron1235, if there's no increment, that doesn't matter, as knowing the number of rounds and the time control lets you easily determine the maximum time.