Hi TL! I'd like to start this discussion off by groaning about the NASL and the lack of good games. Sure there are some epic games to be seen, but watching Grubby get beaten week after week seems a bit boring to me. I've read a bunch of threads by people expressing interest in different formats including something like the NFL which showcases only the best games. But I don't know how well that would work. I think most people agree that the quality of games increases when there is something on the line. In the NASL for example, the few groups near the end where the race for 1st and 2nd was on, the games got significantly more entertaining. I also like the new format for the group stages of the GSL, reverting back to the way the MSL is run with those group games.
Which brings me to my question/point. Why don't more tournaments run competitions in a Swiss style format? For people who follow M:TG, you'll recognize the format of their pro circuit which makes the games exciting and interesting. The next paragraph will explain how the format works, so if you already know, then skip ahead.
How it works:
Take any number of players, say in our case, 256. Each person is paired with another and they play a Bo3. The winners play the winners and the losers play the losers. After 2 rounds, you have 64 0-2, 128 1-1, 64 2-0. This continues on with people with the same (or similar) records playing each other. The number of rounds played is determined by the number of people playing the in the tournament. Since 256 = 2^8, in 8 rounds, you would have one person who was 8-0, 8 people who are 7-1, etc. Then you cut to a top 16 or a top 32. Tie breaks can be broken on many things from games won to games lost to overall win percentage. From here, you seed the top players and continue to single elimination bracket style play.
Pros:
So what makes this format interesting?
1. Every game matters. In those NASL groups where it can down to the end, I'm sure some of the pros wished they had taken prior games more seriously or practiced more. Since this format almost always comes down to tie breaks, it only increases the focus on every game. This increases the overall quality of the games.
2. It weeds out player quickly. The players who continually lose will find themselves having to fight very hard in order to stay in the competition. Swiss tournaments are very harsh on players who lose, and in order to come back they have to continually win. Once again raising the quality of the games.
3. It avoids "groups of death." Remember this recent MSL when Flash, Jaedong, and Bisu were all in one group and only two could make it out? Sure it made the round that much more exciting (MSL groups are a form of a 4 person Swiss round btw) but it also saddened me personally as a Bisu fan when he got knocked out. The Swiss format won't have groups of death and all the pros have a good shot at making it out.
4. It creates exciting match ups. As the tournament progresses, you start to get fewer and fewer pros with flawless records. And every round, there will be matches between players only with perfect records laying everything on the line to stay on top. You'll get to see combinations of all the pros fighting against each other in a randomized manner. You are not restricted to watching the same pros in a single group meet once, then never again until after the group rounds. In a Swiss tournament, you could meet in Round 1 and 7, making for great drama.
Cons:
What are the drawbacks of the Swiss format?
1. Speed. Swiss tournaments narrow the field quickly, meaning tournaments will be shorter and the number of matches to be broadcast will be smaller.
2. Requires Large Player Pools. Swiss Tournaments need large numbers of players to make them more interesting. In a group of 2^5 = 64 players you would only need 5 rounds to narrow it down to the winner.
Conclusion:
I think the Swiss tournament is a highly interesting and intense format that is designed to create great games. In tournaments like the TSL or GSL that don't have a lot of players to start with, the Swiss format may be too quick to be enjoyable. But for open tournaments, it would make them a lot more interesting and give hopefuls a better chance to shine. Imagine the MLG open bracket in a Swiss style. I know the NASL is supposed to have a tournament to fill in the slots for the next season, that would be really cool in a Swiss style too, only the best of the best get in.
Is the Swiss format applicable to Starcraft? Would it make games more exciting? So what do you think?
I like this. Seems it would also make it so more people would get to play more games. I would think for things like the NASL open, where you pay to get in, people would be more motivated to pay to play if they knew they would play more games.
Exactly what I was thinking. The NASL open is 1,024 players. 2^10 = 1024. So the NASL Open would get 10 rounds (exactly the same number of rounds as a bracket btw), but in addition, they get more people all competing for that top spot. In order to win you still have to get no losses, but even if you lose one round, you could still play in another without getting knocked out for sure.
This is essentially the way that competitive debate is handled and I think it would be a great format, sometimes even for smaller tournaments. This is what it was when I debated:
- In a larger tournament with 150+ participants, you would probably have 5-7 preliminary rounds. Less if it's a small pool, but always at least 4.
The first two rounds are against random opponents, and the remaining rounds are "powered" such that the records of your next opponent matches or differs from yours by 1 win or loss.
-After the final preliminary round, you determine who "breaks" into playoffs.
You do this starting with best record and then tie breakers based on criteria like strength of opponent (record).
You decide how long you want playoffs to be (ro128, ro64, ro32, ro16, ro8 and so on) based on the size of the tournament (a ~300 person tournament with a ro64 cut-off is probably standard in the debate world for a big tournament).
And the tournament plays itself out of from there. There are plenty of pros to this format in addition to the ones in the OP.
- Players get plenty of preliminary rounds and they all matter. No one goes home after a single bo3. - It is an extremely fair method in my opinion because it doesn't require seeding outside of the results of the tournament itself. - One upset loss will not prevent you from breaking into playoffs - After the first two preliminary rounds you begin to guarantee really tough matches, with seeding on the line + great potential for rematches in the playoffs.
I really hope more tournaments begin to implement this method for Starcraft 2. My only concern is that for some reason it might be better for participants than for spectators? I know that I never had a problem with it.
of course this should be used a lot more often than it actually is, it's ideal for everyone, not much harder to code than a straight knockout, allows weaker players to get matched up with people closer to their skill level, etc etc
It's obviously a more "fair" system. But the playoff model is a lot more exciting to watch and hypable (if that's a word), as you have to win or be out of the tournament. There is a lot at stake in every single match.
Swiss formats are pretty good in offline tournaments - I'm not a big fan of doing it online.
1) Extra work - hard enough to track players still in the tournament. Extra matches = extra problems that could negatively affect the top competitors.
2) No-shows - sure, if they lose their first two they can still play... but I think no-shows will be rampant unless there's some sort of reward. And seeing how online tournaments generally don't offer prizes to people who don't finish in the top 4 (or 8), may be hard to motivate people to show up for all these games.
3) Confusing - it's hard enough to follow a bracket. Think about what happens when everyone plays every match.
Again, I don't mind it for LANs because if people make a huge effort to come to the league, should be able to get more than a couple of BO3s. But online, to me, makes little sense.
I would've loved to see consolation brackets for the individuals who didn't make Top 2 of their respective groups at Dreamhack though.
I've only ever seen this used in one tournament, and that was a chess tournament I was helping to run with a field of 28 (and we had a couple no-shows, so it was really 26). The plus side is that it doesn't demand a certain number of players. The difficulty is how exactly you work it out. For instance, take a field of 256:
While in some ways it's more ideal, the losers are still losing games and going to be eliminated. There's not a particular difference in length/theory/results between this and a standard single-elim tournament unless you break it off at some point to play "Korean" groups (by which I mean MSL/ODT style), double-elim, or something else - except that everybody plays more games. It would seem primarily useful for establishing a rankings system for the next tournament.
I'm not saying that's necessarily bad, but I'm not sure I see a good application for it. With players being very busy, most of them won't want to take the time to play out a tournament they're already eliminated from. You mention NASL, but I haven't looked at their format in detail.
I am intrigued - I'm trying to brainstorm some ideas because I think it could be made useful.
Quick background: I've been a competitive chess player for almost a decade and played 100s of swiss tournaments.
I've thought about this before, it's different than chess in that chess has draws which makes knockout harder to pair than swiss. However, I think you could do it if you made every round a fixed number of games (e.g. 2 games, not Bo3) and then paired based on those results. You could essentially take that ot mean that 2-0 is like a "win" in chess 1-1 is like a draw, 0-2 is equivalent to a loss. It could create more dynamic formats. There are still the concerns listed and if any organizer has any serious questions about this format he can feel free to PM me and I can work through the possibility of making it work.
It's an interesting idea, but I'm skeptical it will result substantially better events than double elimination with a reasonably high cost.
I think double elimination (MLG, IPL season 1) is my favorite tournament format. It's always exciting to watch a player fall to the losers bracket and fight his way back to the finals, or have two players meet up for a rematch in the losers bracket (MC-Idra at MLG Columbus, for example).
On June 23 2011 06:00 VGhost wrote:
With players being very busy, most of them won't want to take the time to play out a tournament they're already eliminated from.
This is an excellent point as well - most players would drop out once they no longer have a shot at winning, unless they were playing for rankings/seeding in future tournaments/something similar.
On June 23 2011 06:07 MoreFaSho wrote: Quick background: I've been a competitive chess player for almost a decade and played 100s of swiss tournaments.
I've thought about this before, it's different than chess in that chess has draws which makes knockout harder to pair than swiss. However, I think you could do it if you made every round a fixed number of games (e.g. 2 games, not Bo3) and then paired based on those results. You could essentially take that ot mean that 2-0 is like a "win" in chess 1-1 is like a draw, 0-2 is equivalent to a loss. It could create more dynamic formats. There are still the concerns listed and if any organizer has any serious questions about this format he can feel free to PM me and I can work through the possibility of making it work.
It's an interesting idea, but I'm skeptical it will result substantially better events than double elimination with a reasonably high cost.
I know that Chess uses the Swiss format a lot, but I'm aware of it primarily from Magic the Gathering. Those tournaments also use a Swiss Format of say, 8 rounds and then a "cut" or a "break" to a top 8 which is then played in a bracket style.
I do believe however, that this format is advantageous only for tournaments in which there is no previous invitations or seedings. I think its best suited for "open" tournaments. As far as running a tournament with this goes, its not that difficult to code into a program. At the end of each round people submit a result ( 2-0, 2-1, 1-2, 0-2) and the computer spits out the rankings at the end of each round as well as who plays who next. I've seen it work in practice myself. Most of the times ties are broken by your win percentage and then based on your opponents win percentages.
And to the point that people with multiple losses will want to drop, I agree that that factor is there, but Swiss formats can easily account for that. If the lower tier (say people with 3 losses in an 8 round tournament drop), then any player who still wants to play will be given either a "bye" or plays with whoever else wants to play. This lets people who payed to play still get to play for as long as they have opponents, but even if they drop it won't affect the people at the top since they would never play those guys anymore anyway. In fact, its expected people drop. This way at least people get to play at least 2 Bo3s before before leaving.
This is an excellent point as well - most players would drop out once they no longer have a shot at winning, unless they were playing for rankings/seeding in future tournaments/something similar.
As far as professionals are concerned, they would probably all play the games just for the practice and because they would honor their commitment to doing it. The NASL format resulted in some meaningless games involving players from the bottom of their groups, like qxc etc. In some kind of online tournament with amateurs, people would drop but it really is not that big of a deal, you just give a bye or remove players from the pool who aren't there.
EPT
