Winnar
Smash Lord
TL;DR - Swiss is better than round-robin pools in basically every way. If anyone can come up with a reason not to switch to swiss I would love to hear it.
In Chess, which is the most notable example of a game that uses swiss as its primary tournament model, all players are paired together randomly for the first round. They all play their matches, and report the results. Those who won their game in the first round receive 1 point and those who lost receive 0 points. Players with the same score are then paired together randomly for round 2. Repeat the process for round 3, etc. The number of rounds played depends on the number of entrants, but the end result is that players quickly start playing against others closer to their skill level. Closer matches being more enjoyable and desirable, this system practically guarantees that everyone in the tournament will be able to play people at and around their skill level. Additionally, pools are difficult to seed people into, and brackets lose accuracy beyond 4th place. Swiss negates both of these problems with its automatic seeding and its complete tournament placings.
I’ve been asking around and the two biggest complaints I hear about swiss tournaments are that they are less hyped up than bracket matches and that they take too long. The former is easily reconciled by using the top 8 or 16 or even 32 placings of the swiss tournament to seed a double elimination bracket. As for the latter complaint, I feel that this is a huge misconception. I think the main problem is that there is no standardized software for running swiss tournaments. If you actually sit down and look at how long swiss should take versus how long round-robin pools should take, swiss wins almost every time. I wrote out a tremendously long proof, mostly for myself, to show that this is true. I don’t expect anyone but the most dedicated to read it or even look at it, but it is there in case anyone is interested.
[collapse=A Long and Tedious Explanation Provided for Those Interested]There are only two variables to consider when looking at how long a tournament should take: The number of players and the number of setups. The ratio of these two variables (players:setups) determines everything.
When calculating how long pools take, you round up the number of players to the nearest integer for every ratio between the integers (so for example 3.2:1 becomes 4:1). How fast a tournament can run pools is limited to how fast its slowest pool is. As the saying goes, “a chain is only as strong as its weakest link.”
When calculating how fast swiss takes, you round up the number of players to the next even number (so 2.5:1 becomes 4:1). No matter what, each round of swiss will have to be split into two groups of players. One group will play their sets while the other group waits to play theirs, which basically doubles the amount of time each round takes.
For the sake of making this easier to digest let’s say that Tournament X has 100 entrants. Let’s look at a few scenarios:
If the players:setups ratio is 2:1 (aka the optimal ratio)
-Using round robin pools: 10 sets.
To clarify: If everything is run perfectly, pools should take the same amount of time as 2 players running 10 long best of 3 sets in a row. Think of a set as a length of time, like 15-20 minutes.
[collapse=Explanation]There is a lot of leeway here on how this can be done, but what I might do is two rounds of pools. First round would be six players in each and top 3 get out, which weeds it down to 50 players left. Second round would be five players in each and top 3 get out, which brings the players left to 30. This is not an optimized pools setup, but for the sake of this example that’s okay. The math works out that within the first round pools, 15 games need to be played and they will be played at a rate of 3 games at a time. This means that ideally round one of pools will take the time of 5 sets if everyone is cooperative and perfectly organized/instructed on who plays who. Round two of pools has 10 games that can be played at
2 games at a time. This means 5 more sets, so pools takes 10 sets before you can start brackets.[/collapse]
-Using swiss style brackets: 7 sets
[collapse=Explanation]The number of rounds you need in a swiss bracket depends only on the number of players. You can use the following formula to determine the rounds you need.
[collapse=Formula and Example]2^(n-1) < Number of players <= 2^n, where n = number of rounds needed
So for example if you have 14 players then n = 4, because 2^3 = 8 and 2^4 = 16 and 14 is between those two numbers.[/collapse]
Anyway, you need 7 rounds to accommodate 100 players. If they all run simultaneously then swiss only takes 7 sets. You can then set up your bracket using the swiss results for seeding.[/collapse]
If the players:setups ratio is 3:1
-Using round robin pools: 13 sets
[collapse=Explanation]Same setup as before. Two rounds of pools: 6 person per pool first round and 5 person per pool second round, top three players make it out of both rounds of pools. This time there are only 2 setups available per pool, so 8 sets need to be played for first round of pools. As before, the second round of pools will take the time of 5 more sets. Therefore, pools takes 13 sets if everything is ran flawlessly.[/collapse]
-Using swiss style brackets: 14 sets
[collapse=Explanation]As explained above, each round will need to be divided into two groups of players: Those who play their sets first and those who play their sets second. This means each round of swiss essentially takes two sets to finish each round. Twice 7 is 14, so it takes 14 sets to run swiss with this ratio.[/collapse]
If the players:setups ratio is 4:1
-Using round robin pools: 17 sets
[collapse=Explanation]Here things get a little complicated. Using the same system of two rounds of pools, two-thirds of the pools in the first round will use 2 TV’s and take 8 sets. The other third will only have a single TV in each pool and will take 15 sets, sort of. After 8 sets, pools with 2 TV’s will finish and so the next 7 sets that need to be finished on the 1 TV pools will faster. These 7 sets will only take 4 sets time. So the first round of pools will take about 12 sets to finish. This problem is avoided in round 2 pools because only 50 players will make it to round 2, so setups are plentiful at this point. Second round pools takes 5 sets, so both pools together take about 17 sets to completely finish.[/collapse]
-Using swiss style brackets: 14 sets
[collapse=Explanation]See 3:1 swiss setup. Do note, however, that 3:1 will probably run slightly faster than 4:1. Statistically speaking, there’s less likelihood that your second group of players will have a set that takes a long time. That said, the difference in time is probably not going to be very significant.[/collapse]
If the players:setups ratio is 5:1
-Using round robin pools: 20 sets
[collapse=Explanation]I am going to assume for the first pool that this will take about the same amount of time as running it with a ratio of 6:1. The reason for this is that only 1/6 of the pools will have two TV’s, and by the time all the pools have access to 2 TV’s it will probably not save them any time anyway. So then all 15 sets will take the full time of 15 sets. Round 2 of pools will take the usual 5 sets, so pools as a whole will take 20 sets to finish.[/collapse]
-Using swiss style brackets: 24 sets
[collapse=Explanation]7*3=24; see 3:1 swiss explanation.[/collapse]
If the players:setups ratio is 6:1
-Using round robin pools: 20 sets
See 5:1 explanation
-Using swiss style brackets: 24 sets
See 5:1 explanation
Oh, and here are two examples of what results will look like if you have 8 players and 16 players:
[collapse=8 players]
round 1
4x 1-0
4x 0-1
round 2
2x 2-0
4x 1-1
2x 0-1
round 3
1x 3-0
3x 2-1
3x 1-2
1x 0-3[/collapse]
[collapse=16 players]
round 1
8x 1-0
8x 0-1
round 2
4x 2-0
12x 1-1
4x 0-2
round 3
2x 3-0
6x 2-1
6x 1-2
2x 0-3
round 4
1x 4-0
4x 3-1
6x 2-2
4x 1-3
1x 0-4[/collapse]
[/collapse]
So, whenever your player to setup ratio is 4:1 or better, swiss will be faster. This is even after making some very generous assumptions on how quickly pools can be run. As I said earlier a tournament is only as fast as its slowest pool, and it’s difficult to have 15 pools without one or two of them being run inefficiently.
In any case, if you have a bad player to setup ratio then pools is probably going to be faster. I say that, but even if the ratio is close to 6:1 or so then swiss might still be a good option. Swiss gives such a better measure of accuracy for your tournament results since it measures the entire tournament player base at once. With more accurate results for an entire tournament entry, we as a community will be able to set up a unified ranking system. Chess uses swiss and has a highly successful ranking structure. I need to do some more research as to how chess's elo ranking system works, but that's another discussion on its own. If you are an expert on elo ranking then I would ask that you please either point me in the right direction for learning about it or just share your knowledge here.
Here’s a quick breakdown:
Pros
Cons
As I said before, making the software is trivial and I can make a makeshift excel visual basic macro that will get the job done perfectly fine until someone with better coding experience steps up and makes something more legit. Getting setups can be a challenge at first, but if the TO places a high emphasis on getting setups to tournaments (via incentives like no venue fee or by simply explaining how important it is for a tournament to have setups) then that goal is easily achievable.
Oh, I guess I should mention that tiebreakers can be a little confusing. Ties tend to happen in swiss when the number of entrants is midway between a power of 2. This is an interesting issue, though, because there are a few ways to solve tiebreakers. One way is to use the statistics automatically generated to tell who in the tied group had the hardest matches. These will be exported in the software anyway, and they will be an accurate judge of who fought the hardest within the tied group, but it's a bit anticlimactic. The other way (the more fun/hype way) is to simply have the tied players play out their tiebreaker matches against each other. For a 2 person tiebreaker, it would just be one set; for a 3 person tie-breaker, a round robin; and anything larger gets a single elimination bracket seeded by the statistics. Tiebreaker matches could end up being more hype than most bracket matches.
If anyone can think of any other problems or concerns with running swiss, I’d love to hear them. Is there any reason why we shouldn’t switch to swiss?
In Chess, which is the most notable example of a game that uses swiss as its primary tournament model, all players are paired together randomly for the first round. They all play their matches, and report the results. Those who won their game in the first round receive 1 point and those who lost receive 0 points. Players with the same score are then paired together randomly for round 2. Repeat the process for round 3, etc. The number of rounds played depends on the number of entrants, but the end result is that players quickly start playing against others closer to their skill level. Closer matches being more enjoyable and desirable, this system practically guarantees that everyone in the tournament will be able to play people at and around their skill level. Additionally, pools are difficult to seed people into, and brackets lose accuracy beyond 4th place. Swiss negates both of these problems with its automatic seeding and its complete tournament placings.
I’ve been asking around and the two biggest complaints I hear about swiss tournaments are that they are less hyped up than bracket matches and that they take too long. The former is easily reconciled by using the top 8 or 16 or even 32 placings of the swiss tournament to seed a double elimination bracket. As for the latter complaint, I feel that this is a huge misconception. I think the main problem is that there is no standardized software for running swiss tournaments. If you actually sit down and look at how long swiss should take versus how long round-robin pools should take, swiss wins almost every time. I wrote out a tremendously long proof, mostly for myself, to show that this is true. I don’t expect anyone but the most dedicated to read it or even look at it, but it is there in case anyone is interested.
[collapse=A Long and Tedious Explanation Provided for Those Interested]There are only two variables to consider when looking at how long a tournament should take: The number of players and the number of setups. The ratio of these two variables (players:setups) determines everything.
When calculating how long pools take, you round up the number of players to the nearest integer for every ratio between the integers (so for example 3.2:1 becomes 4:1). How fast a tournament can run pools is limited to how fast its slowest pool is. As the saying goes, “a chain is only as strong as its weakest link.”
When calculating how fast swiss takes, you round up the number of players to the next even number (so 2.5:1 becomes 4:1). No matter what, each round of swiss will have to be split into two groups of players. One group will play their sets while the other group waits to play theirs, which basically doubles the amount of time each round takes.
For the sake of making this easier to digest let’s say that Tournament X has 100 entrants. Let’s look at a few scenarios:
If the players:setups ratio is 2:1 (aka the optimal ratio)
-Using round robin pools: 10 sets.
To clarify: If everything is run perfectly, pools should take the same amount of time as 2 players running 10 long best of 3 sets in a row. Think of a set as a length of time, like 15-20 minutes.
[collapse=Explanation]There is a lot of leeway here on how this can be done, but what I might do is two rounds of pools. First round would be six players in each and top 3 get out, which weeds it down to 50 players left. Second round would be five players in each and top 3 get out, which brings the players left to 30. This is not an optimized pools setup, but for the sake of this example that’s okay. The math works out that within the first round pools, 15 games need to be played and they will be played at a rate of 3 games at a time. This means that ideally round one of pools will take the time of 5 sets if everyone is cooperative and perfectly organized/instructed on who plays who. Round two of pools has 10 games that can be played at
2 games at a time. This means 5 more sets, so pools takes 10 sets before you can start brackets.[/collapse]
-Using swiss style brackets: 7 sets
[collapse=Explanation]The number of rounds you need in a swiss bracket depends only on the number of players. You can use the following formula to determine the rounds you need.
[collapse=Formula and Example]2^(n-1) < Number of players <= 2^n, where n = number of rounds needed
So for example if you have 14 players then n = 4, because 2^3 = 8 and 2^4 = 16 and 14 is between those two numbers.[/collapse]
Anyway, you need 7 rounds to accommodate 100 players. If they all run simultaneously then swiss only takes 7 sets. You can then set up your bracket using the swiss results for seeding.[/collapse]
If the players:setups ratio is 3:1
-Using round robin pools: 13 sets
[collapse=Explanation]Same setup as before. Two rounds of pools: 6 person per pool first round and 5 person per pool second round, top three players make it out of both rounds of pools. This time there are only 2 setups available per pool, so 8 sets need to be played for first round of pools. As before, the second round of pools will take the time of 5 more sets. Therefore, pools takes 13 sets if everything is ran flawlessly.[/collapse]
-Using swiss style brackets: 14 sets
[collapse=Explanation]As explained above, each round will need to be divided into two groups of players: Those who play their sets first and those who play their sets second. This means each round of swiss essentially takes two sets to finish each round. Twice 7 is 14, so it takes 14 sets to run swiss with this ratio.[/collapse]
If the players:setups ratio is 4:1
-Using round robin pools: 17 sets
[collapse=Explanation]Here things get a little complicated. Using the same system of two rounds of pools, two-thirds of the pools in the first round will use 2 TV’s and take 8 sets. The other third will only have a single TV in each pool and will take 15 sets, sort of. After 8 sets, pools with 2 TV’s will finish and so the next 7 sets that need to be finished on the 1 TV pools will faster. These 7 sets will only take 4 sets time. So the first round of pools will take about 12 sets to finish. This problem is avoided in round 2 pools because only 50 players will make it to round 2, so setups are plentiful at this point. Second round pools takes 5 sets, so both pools together take about 17 sets to completely finish.[/collapse]
-Using swiss style brackets: 14 sets
[collapse=Explanation]See 3:1 swiss setup. Do note, however, that 3:1 will probably run slightly faster than 4:1. Statistically speaking, there’s less likelihood that your second group of players will have a set that takes a long time. That said, the difference in time is probably not going to be very significant.[/collapse]
If the players:setups ratio is 5:1
-Using round robin pools: 20 sets
[collapse=Explanation]I am going to assume for the first pool that this will take about the same amount of time as running it with a ratio of 6:1. The reason for this is that only 1/6 of the pools will have two TV’s, and by the time all the pools have access to 2 TV’s it will probably not save them any time anyway. So then all 15 sets will take the full time of 15 sets. Round 2 of pools will take the usual 5 sets, so pools as a whole will take 20 sets to finish.[/collapse]
-Using swiss style brackets: 24 sets
[collapse=Explanation]7*3=24; see 3:1 swiss explanation.[/collapse]
If the players:setups ratio is 6:1
-Using round robin pools: 20 sets
See 5:1 explanation
-Using swiss style brackets: 24 sets
See 5:1 explanation
Oh, and here are two examples of what results will look like if you have 8 players and 16 players:
[collapse=8 players]
round 1
4x 1-0
4x 0-1
round 2
2x 2-0
4x 1-1
2x 0-1
round 3
1x 3-0
3x 2-1
3x 1-2
1x 0-3[/collapse]
[collapse=16 players]
round 1
8x 1-0
8x 0-1
round 2
4x 2-0
12x 1-1
4x 0-2
round 3
2x 3-0
6x 2-1
6x 1-2
2x 0-3
round 4
1x 4-0
4x 3-1
6x 2-2
4x 1-3
1x 0-4[/collapse]
[/collapse]
So, whenever your player to setup ratio is 4:1 or better, swiss will be faster. This is even after making some very generous assumptions on how quickly pools can be run. As I said earlier a tournament is only as fast as its slowest pool, and it’s difficult to have 15 pools without one or two of them being run inefficiently.
In any case, if you have a bad player to setup ratio then pools is probably going to be faster. I say that, but even if the ratio is close to 6:1 or so then swiss might still be a good option. Swiss gives such a better measure of accuracy for your tournament results since it measures the entire tournament player base at once. With more accurate results for an entire tournament entry, we as a community will be able to set up a unified ranking system. Chess uses swiss and has a highly successful ranking structure. I need to do some more research as to how chess's elo ranking system works, but that's another discussion on its own. If you are an expert on elo ranking then I would ask that you please either point me in the right direction for learning about it or just share your knowledge here.
Here’s a quick breakdown:
Pros
- It lets more players get more matches
- It’s faster even despite the first point
- It holds equal appeal players of all skill levels
- It seeds better for brackets
- It’s easier to run a swiss tournament
- It allows for a unified ranking system similar to chess’s elo system
Cons
- It requires a higher emphasis on getting setups to tournaments
- It needs software for best results
- It suffers from some misconceptions, so introducing it could be a bit of a gamble for TO’s
- It might be more difficult to enforce rules and provide guidance for new players than pools (no pool leaders)
As I said before, making the software is trivial and I can make a makeshift excel visual basic macro that will get the job done perfectly fine until someone with better coding experience steps up and makes something more legit. Getting setups can be a challenge at first, but if the TO places a high emphasis on getting setups to tournaments (via incentives like no venue fee or by simply explaining how important it is for a tournament to have setups) then that goal is easily achievable.
Oh, I guess I should mention that tiebreakers can be a little confusing. Ties tend to happen in swiss when the number of entrants is midway between a power of 2. This is an interesting issue, though, because there are a few ways to solve tiebreakers. One way is to use the statistics automatically generated to tell who in the tied group had the hardest matches. These will be exported in the software anyway, and they will be an accurate judge of who fought the hardest within the tied group, but it's a bit anticlimactic. The other way (the more fun/hype way) is to simply have the tied players play out their tiebreaker matches against each other. For a 2 person tiebreaker, it would just be one set; for a 3 person tie-breaker, a round robin; and anything larger gets a single elimination bracket seeded by the statistics. Tiebreaker matches could end up being more hype than most bracket matches.
If anyone can think of any other problems or concerns with running swiss, I’d love to hear them. Is there any reason why we shouldn’t switch to swiss?