Round Robin and Swiss System tournaments: an alternative to double elim

[Rapid_Assassin]

Recently, I was at a small local tournament that I did very well at. Granted, the results would have been different if there weren't people who left early, and if there had been tie breakers. And even so, I still wouldn't have been dead last (like I often am). The main difference between this and every other tournament I've ever been to was that it was entirely round robin. Everyone played vs. everyone else.

There are benefits to a round robin system regardless of your skill level:

1. For new people starting out, most double elim tournaments are seeded so that they end up playing vs. the best one there first round, and then the best one in the "bottom half" second round. This can be discouraging for someone who is trying to learn and get better. They still wouldn't have a chance to win the tournament, but it is better for them to get experience vs. multiple opponents, not the same people that they always lose to every tournament.

2. For average players (by tournament standards), the round robin will generally place them higher than a double elim bracket, proving basically the same thing that my small local tournament did: they aren't the worst players there. Also, it gives the same benefit of playing vs. multiple opponents that the double elim bracket generally won't give.

3. For pros, what better way is there to prove that you dominate at the game than to beat every single person there?

Of course, there does need to be some tweaking to some formats of round robin that I've seen. Only winning a set vs. someone should count. If you won 1 match and still lost the set, it should count the same as if you lost both matches. Also, if there are any ties between the top 4 players (only ones concerned with winning money anyway) there should be a tie breaker set played between them. And any ties between players ranked below 4th should be noted in the results.

Unfortunately, round robin takes a long time, and isn't realistic if it's a large tournament. A better solution for larger tournaments would be the Swiss System. A Swiss tournament will have the same benefits as a round robin, without an unrealistic number of matches that needs to be played. In a larger tournament, Swiss system will take about as long as double elimination will, produce more accurate results, and provide the same benefits as round robin for non-pro players.

For those unfamiliar with tournament theory:

http://en.wikipedia.org/wiki/Swiss_system_tournament
http://en.wikipedia.org/wiki/Double-elimination_tournament
http://en.wikipedia.org/wiki/Round-robin_tournament

Discuss

 

[Blackadder]

I'm tired, and in a "Blackadder spamming session" mood.
*Spaaaaaaam*
Okay, I'm happy now.

On topic:
As you said, smaller tournys=Round robin
Large ones=Swiss system.
Elim is still just "TEH BOMBZ" though. It's classic, and has an intimidating name to boot! I think the people running the tourny should just decide how it'll work depending on what the situation calls for at the time.

^^
people, if non of that makes sense it's cause I stayed up all night, and don't even recall what my left nipples looks like anymore.
Go easy on me.

 

[AlphaZealot]

There is a problem with RR that you can't see on paper, and that is player fatigue. This will happen in any tournament format, but, when you have a tournament like Melee-FC Diamond, that has 256 people, best of 3 matches, 3 rounds of pools, then a bracket, your looking at people having already played 60+games before you even start to get toward the best players.

FC was origanalyl suppose to be a round robin for the top 8 players (after a bracket). I can tell you from being there none of the top 8 players would have been playing their best in this format, and most struggled to get through the double elimination portion, after all many of them were up until 4-5 am playing in the Smash 64 tournament the night before.

 

[Rapid_Assassin]

There is a problem with RR that you can't see on paper, and that is player fatigue. This will happen in any tournament format, but, when you have a tournament like Melee-FC Diamond, that has 256 people, best of 3 matches, 3 rounds of pools, then a bracket, your looking at people having already played 60+games before you even start to get toward the best players.

That's why I mentioned Swiss format for large tournaments (anything over 30-50, depending on the number of setups). For example, in a Swiss format tournament of 100 people, it would take 7 rounds to determine a winner, plus an additional round if there's a tie for the top 4. In a double elimination bracket of 100 people, it takes 8 rounds to determine a winner. That is the top player playing vs. 8 people in both tournament formats.

 

[pockyD]

rr/swiss = better accuracy but you lose the excitement of elimination, plus a lot more matches usually have to be played (time issue)

 

[Rapid_Assassin]

Also, one thing I forgot to mention about Swiss format is that it lessens or eliminates the most common bracket complaints.

1. Two pros playing first round. Say Player A beats Player B first round, both of them are pros. Both players beat every single other person they have to fight. Player A wins the tournament. Player B gets 2nd (assuming that there are no ties). Since there's no true "loser's bracket" someone else who has to fight Player B can easily get 3rd.

2. "Hell brackets" Not an issue, since winners are placed with winners. There is a possibility of one person getting 4th having a harder set of people to play against than another person getting 4th. Only way to completely eliminate this is a full round robin, which as I mentioned is unrealistic for large tournaments.

3. People playing against their own crew members. First round should be seeded by location in this format, not by skill. This eliminates having to play against someone from your crew, town, etc..

 

[Dark Sonic]

I think you don't quite understand what Rapid Assassin is saying. The swiss system takes the same number of rounds so as long as you have enough TVs (1 to every 2 players is ideal, but 1 to every 4 works well too) it won't take much longer than a double elimination tourney and everyone will have more matches. I like the idea of an elimination tourney between the top players near the end to determine a clear winner. Although the pools system works well in addition to the double elimination bracket, but that has it's own flaws too.

 

[pockyD]

"rounds" and "sets are not equal"

if you have 64 entrants, your standard double elimination bracket will play 126-127 sets

a swiss system would need 6 rounds, meaning there will be 192 sets played (~50% increase)

with even more entrants, it scales up even more

not saying it's a bad idea, but counting the number of rounds is inaccurate

 

[Rapid_Assassin]

Elimination between say the top 4 (plus any that tied with them) combined with Swiss is a very good idea. Seed the final elimination bracket so that people that already played each other don't do so again (if avoidable). This will allow for the same excitement in elimination tournaments for anyone who would be concerned with winning money.

Also, while more sets are played, it'd still take approximately the same amount of time. The number of TVs doesn't change. In elimination, usually when over 1/2 of the people are eliminated, there's a lot of free TVs that aren't being used for the tournament anymore. If the tournament is Swiss, all TVs are used throughout the whole tournament, so the extra matches being played when compared with elimination are played in later rounds, when TVs would have been free otherwise.

 

[knightpraetor]

wow i'd never heard of the swiss system before but i like it a lot..it appears to be more accurate, more fun for new players..less tiring on players as a whole..

 

[thespymachine]

Being inspired by the revitalization of rankings (MIOM's Top 100, ssbpd being picked up again, and SWF's new player rankings), I started working on my ELO ratings that I did for Vegas Melee while I lived there.
Doing what I do, I did some more research into Elo and the competitive chess scene. I read more into Swiss and how they used it, with a bunch of variations on it.
After doing a bunch of experimenting and reevaluating on my Elo stuff, I came to the conclusion that ELO and Elimination-Brackets do not mix (since Elo is most accurate when players have played the same amount).
So, in addition to some Elo stuff I'm doing for Vegas right now, I've been working on a Swiss-style tournament that is modified to embrace elements of both elimination (hype) and round robin (accuracy) brackets.
I call it Super-Swiss.

1. Number of Rounds

The number of rounds is determined by the number of entrants (like RR and S/DE are as well).
We need enough rounds to determine a top player in the bracket. I consider this as many rounds needed to get only one player without a loss (or, the winner of Winner's Finals). Typically, it seems, most Swiss tournaments are ran to this point which is quick and hardly different from single-elimination, but has little accuracy (especially if seeding isn't done, which is seemingly hardly ever done - see #2), so: I give the rest of the competition a chance at that player/top spot - two more rounds.

​

First, how many rounds to get to WFs, dependent on # of players:

3-4 players = 4-player bracket = 2 rounds until WFs is completed
5-8 players = 8-player bracket = 3 rounds until WFs is completed
9-16 = 16 bracket = 4 rounds
17-32 = 32 = 5
33-64 = 64 = 6
etc, etc
​

Second, add two rounds to that:

3-4 players = 2 rounds + 2 more = 4 (at this point, you're playing RR with an additional round)
5-8 players = 3 rounds + 2 more = 5
9-16 = 6 total rounds
17-32 = 7 total rounds
33-64 = 8
etc, etc
​

So, Super-Swiss (SS) ends up being quicker than RR when there are more than 6 players.
​

2. Seeded

Like an elimination bracket, first round is highest seed plays vs lowest seed, second highest vs second lowest, etc.
And with each round after, besides the last 2 (see #3), seeding is used in the same way with players with the same record - in all the players who are 2-0, the highest seed plays the lowest seed, etc (the only exception being in cases where you get an odd # of players with the same record, when the highest seed will drop down to players with a slightly worse record and play their lowest seeded player).
What this seeding does is mitigate the chances that the better players will face each other before they need to. And, as mentioned before, it seems that a lot of Swiss tournaments aren't run this way, which could potentially get your top two players playing in first round.

In combination with Swiss' main rule (people with similar records play each other), seeding makes players in similar skill play each other and makes the whole tournament play out like an elimination bracket without any eliminations.
​

3. Tiebreak

The tiebreak method I use is a variation of the Median-Buchholz System where a player's tiebreak # is the number of wins of the players they faced, disregarding the highest and lowest records (to mitigate inaccuracies in the earlier rounds), divided by # of games played. So: r = total wins of opponents (except the highest and lowest records), g = # of games played

tiebreak # = r/g​

In the last two rounds, instead of using the seedings of the players, I use their tiebreak #s - where the it's players with the closest tiebreak numbers play each other. [Unlike the earlier rounds with the seeding process, the player with the lowest tiebreak number, in an odd numbered group of players with the same record, plays someone with the next closest record; 3 players are 3-0, one has a tb# of 2, one with 2, and the other with 1 - the player with the tb# of 1 would play a player with a 2-1 record, and as close to their tb# as possible].

This builds even further on Swiss' intention on getting players with similar skill levels. Where seeding helps establish an initial separation of player skill via past performances, players with similar records takes into account who's playing similarly on that day, while tiebreak considers who've the players have played for a more accurate equal-skilled matches.
​

4. No Repeat Matches Until Last Round

A potential big flaw of Swiss is allowing for the same players to play against each other over and over because they are constantly finding themselves with the same record.
SS doesn't allow repeats until the last round. The wider range of players played gives a better judgement to a players skill and doesn't force players into the same MUs over and over (pluses for the TO and the players).
And, in the end, it allows for hype, where you will be getting (very very likely) to two most evenly matched games you'll play through the whole tournament.

Taking a bit from RR, you get more matches from more players, which equals accuracy, slower decay of player interest, and hype sets the last round.
​

5. Consolation Round?

When the last round is finished, there is potential for the players who faced each other in the last round to be tied in sets vs each other. When this happens, the TO can call for a FINAL round where those same players face off against each other again - continuing hype.
This isn't needed, since Record and Tiebreak will be used to break ties - but even then there could be ties, which don't NEED to be broken unless the TO/players want to.
​

6. Byes

Byes are difficult to figure out, since they don't work the same as they do in elimination brackets.
Who I give a bye to is decided by:

-Someone who hasn't had a bye yet in the tournament

^Worst record

^Lowest tiebreak

^Lowest seed
​

This insures that, since byes actually hurt your chances to win, 1) no one gets a bye twice, and 2) the people least likely to win (via record, tiebreak, AND seed) are given them.
​

Summary by Round
First rounds: Best of 3

1. Same/close record
2. Haven't played
3. Seeding​

Third to last round: Best of 5

1. Same/close record
2. Haven't played
3. Seeding​

Second to last: Best of 5

1. Same/close record
2. Haven't played
3. Tiebreak​

Last round: Best of 5

1. Same/close record
2. Tiebreak

​

Overall, Swiss is faster than RR and more accurate than DE. SS isn't run as fast as bracket pools, but can be drastically faster than RR.

 

[SAUS]

My biggest gripe with swiss or round-robin is that there are no true 'grand finals'.

1. Two pros playing first round. Say Player A beats Player B first round, both of them are pros. Both players beat every single other person they have to fight. Player A wins the tournament. Player B gets 2nd (assuming that there are no ties). Since there's no true "loser's bracket" someone else who has to fight Player B can easily get 3rd.

In this example, essentially, first round is the grand finals. I just don't like the way it is. A huge thing about double-elimination is the hype you get. No one is playing serious games anymore (except maybe money-matches), so everyone watches the grand finals, and the hype and atmosphere is maximised.

I remember there used to be biweekly's for barlw in my area and they were pretty fun. Then someone decided we should try swiss, and it was really, really boring. There were no grand finals so no one watched any of the matches unless there was no TV for them. It was overall just not nearly as awesome of an experience.

Elimination between say the top 4 (plus any that tied with them) combined with Swiss is a very good idea. Seed the final elimination bracket so that people that already played each other don't do so again (if avoidable). This will allow for the same excitement in elimination tournaments for anyone who would be concerned with winning money.

This is actually a pretty cool solution, but isn't this kind of already how it works (sort of)? In all the tournaments I've been to, it's pools (round-robin), and then the top players from there go to brackets. When I was a new player, I got my ass whooped, but I still play the 8 people in my pool. To me, that was enough to get some experience and see how good I was at the game (terrible lol, I lost all the games). After that, you just run around getting in on the friendlies to play against more people. That still doesn't mean that it is better the way it is now compared to the way you describe here, I just feel that changing it will make that much of a difference.

 

[JU4N]

Pools then bracket seems optimal. Pools shouldn't be too stressful for top players, and new players get more matches. Then a bracket at the end of the tournament promotes hype.

This is all assuming that time permits, of course.

 

[TheCrimsonBlur]

Swiss sucks

 

[thespymachine]

Swiss sucks

at being horrible? lol

 

[JKJ]

I just like RR pools. Thats how it used to be, and I still feel the best about that. Bracket pools don't really prove who the best in the pool is, just who was the best in conjunction with having an easier bracket is. I believe RR pools, then double elimination bracket. Everyone starts in winners in bracket. I think that starting in losers out of bracket pools is terrible.

 

[thespymachine]

I just like RR pools. Thats how it used to be, and I still feel the best about that. Bracket pools don't really prove who the best in the pool is, just who was the best in conjunction with having an easier bracket is. I believe RR pools, then double elimination bracket. Everyone starts in winners in bracket. I think that starting in losers out of bracket pools is terrible.

Yeah, it seems TOs sacrifice accuracy/fairness for speed.
You get way more accuracy/fairness from round-robin and swiss, than you do from any elimination bracket. (Not to bash the players in Apex 2014's Top 8, but I'm quite confident that there would have been a different line-up if there were RR pools instead of bracket.)

And I dislike the final bracket having a winners and losers side automatically.

 

[Deader Survival]

There's one thing that NONE of you have thought of. Ratings.

Sure, as a TO, you can just randomize players or have them play some matches. But that eliminates the point of a tournament. As a TO, I'll keep in mind everything that you guys have said, but, one issue, how do I rank people when most of the people that will be at my event will be unranked to the general population?

That's why I enjoy the simplicity of DE into DE. Players don't have to wonder about who they'll be going against next, and it allows for a faster tourney in general. Plus I don't have to sit around for a couple hours trying to figure out the fairest bracket. I randomize it and let everyone roll.

Sure, it's not 100% infallible but if you're there to win, you should stop. The point of some tourneys is to have fun. Plus, the best players will be at least in the latter rounds.

And for the new players, go find a ladder to play on. Ladders are fairer to new players than tournaments are.

Anyway, have fun, feel free to counter me if you have something I forgot. ^-^