Helpful spreadsheet for 8 player non-elimination tournament.

Luke Burrage

Hi fellow geeks,

After taking part in a load of different style of tournaments, my friends and I were getting annoyed with the different systems. A straight knockout was a annoying because the second best team could be eliminated in the first round. Using "seeding" and group stages was annoying because the good teams (our teams) would have to play through a load of easy games before doing anything interesting. We tried a variation on the "Swiss System" but found it lasted a long time and didn't suit only having one playing area.

So I put together a variation of a tournament, based in part on the Swiss System and part on other ideas. It is probably too complex to work just using pen and paper, but these days everyone has a laptop so I thought I'd make a spreadsheet that a tournament organizer can easily use on any platform, hence the .xls format. Please download it, follow the instructions, play with it, test out different results and tell me what you think! For now I just have an 8 team spreadsheet ready to go, but it can be extended to 12, 14, 16, etc teams without too much trouble.

8 teams 3 games each tournament version 004.xls

Some more information:
1. The tournament is for games or sports where points or goals determine the winner. For games like chess (simple win or lose) it still works but by the end there is a clear winner and clear loser, but 3 teams will be placed joint second and 3 teams joint 5th. A 4th round would be needed to resolve this.

2. Ties sort of break the system, so games like Scrabble or football don't work so well. The tournament system was designed for games like volleyball where a winner is inevitable.

3. The rankings are based first on wins and losses. However, unlike the swiss system, you do not simply get one point for winning. Instead it looks for the highest score of the tournament so far and uses three times that as the value you get for winning a game. Then it looks at the point difference between the two scores and modifies the ranking by that much. In this way, a team who wins two games in a row by just one point will be rated higher than a team who wins by 20 points one game and loses by one point the next game.

4. When the games are played in the order dictated, the final game will decide the winner. Unlike a normal 8 team 3 round knockout, the loser of the final game might not be the second place finisher.

5. A friend made a macro to do the select and sort with a single click but it didn't work across Excel, Open Office and Neo Office.


I hope you find it interesting or useful or at least give some helpful feedback. I will not respond to anyone moaning about how limited it is. I KNOW that, but this has come in handy a few times to make organizing simple, fair tournaments, and I just want to make it better.

Apreche

Your solution fundamentally changes the nature of the games themselves since you take into account scores instead of just wins and losses. For pretty much every sport or game, the rationale is that who wins and who loses is what is important. What the final score happened to be is not important. Thus, every player/team is encouraged to score as many points as possible to ensure victory, which is all that matters.

In this way, a team who wins two games in a row by just one point will be rated higher than a team who wins by 20 points one game and by one point the next game.

This incentive fundamentally changes the game being played, as it discourages the team in the lead from scoring too many points. Teams in the lead would switch to 100% defensive strategies rather than continue trying to score. This breaks the first rule of tournament structures in that they should not modify the game being played in any way.

Also, good luck trying to get people to understand how this works. A simple and familiar tournament structure is usually much better than something weird, even if it is flawed. Otherwise, people will complain and be confused. If you try something like this where there is a prize on the line, be prepared for people to treat you like you cheated them out of the prize with your weird tournament.

There are any number of existing, simple, tried and true, tournament structures available. Your justification for a new, more complicated structure, which has a negative effect on the games themselves, is pretty weak. If you don't want to seed a bracket, just do a loser's bracket. If you don't like the swiss system, just do a round robin. If a complete round-robin will take too long, just do a limited randomized round robin that will be used to seed a very small elimination bracket.

lackofcheese

We tried a variation on the "Swiss System" but found it lasted a long time and didn't suit only having one playing area.

The Swiss System takes just as many rounds as your system.
You really haven't made a clear case for
1) What's wrong with the Swiss System
2) What makes your system better than it.

Luke Burrage

Your solution fundamentally changes the nature of the games themselves since you take into account scores instead of just wins and losses. For pretty much every sport or game, the rationale is that who wins and who loses is what is important. What the final score happened to be is not important. Thus, every player/team is encouraged to score as many points as possible to ensure victory, which is all that matters.

I think I didn't write this clearly enough. The rating has nothing to do with the final results of the tournament. All it does is rank the teams with who they will play in the next round. Winning is the only thing that determines the final winner of the tournament, as the only team that wins three games in a row will win. Just like a straight knockout tournament.

In this way, a team who wins two games in a row by just one point will be rated higher than a team who wins by 20 points one game and by one point the next game.

This incentive fundamentally changes the game being played, as it discourages the team in the lead from scoring too many points. Teams in the lead would switch to 100% defensive strategies rather than continue trying to score. This breaks the first rule of tournament structures in that they should not modify the game being played in any way.

I actually made a big typo here. What I meant to say is:

"In this way, a team who wins two games in a row by just one point will be rated higher than a team who wins by 20 points one game and loses by one point the next game."

In other words, the winning is ALWAYS the most important thing to do, and the only thing to do to win. The mechanic for sorting out who plays who is so that at the end of the tournament, teams are playing with other teams of equal skill. If you lose by lots in the first two rounds, you'll be matched with someone who has also lost by lots in the first two rounds.

Also, good luck trying to get people to understand how this works. A simple and familiar tournament structure is usually much better than something weird, even if it is flawed. Otherwise, people will complain and be confused. If you try something like this where there is a prize on the line, be prepared for people to treat you like you cheated them out of the prize with your weird tournament.

There is nothing weird about the tournament, it just uses a non-random way of assigning teams in the next round. For the winning team it is indistinguishable from a straight knockout tournament, but it lets other teams play more games and has fairer matches near the end. Like I said, it isn't the best system for all situations, but it is better than randomized pairings in later rounds, and takes less time than a full round robin, while giving each team an equal number of games.

Hope that clears things up.

Apreche

I think I didn't write this clearly enough. The rating has nothing to do with the final results of the tournament. All it does is rank the teams with who they will play in the next round. Winning is the only thing that determines the final winner of the tournament, as the only team that wins three games in a row will win. Just like a straight knockout tournament.

Even so, having better ratings will change who your opponent will be in future rounds. Teams will be encouraged to either run up the score, or crank up the defense, in order to get weaker opponents in future rounds. Rank should only be determined by number of wins and losses. The in-game strategies need to be completely untouched by tournament structure. While playing a game, having more points than the opponent should be the only thing on the mind of the participants. Even a small effect on that tarnishes the purity factor.

There is nothing weird about the tournament, it just uses a non-random way of assigning teams in the next round. For the winning team it is indistinguishable from a straight knockout tournament, but it lets other teams play more games and has fairer matches near the end. Like I said, it isn't the best system for all situations, but it is better than randomized pairings in later rounds, and takes less time than a full round robin, while giving each team an equal number of games.

It's weird in that it is not something the majority of people are familiar with. Even if it did produce a better tournament, the trouble it would cause from a PR perspective would make it difficult. The participants in the tournament are likely to complain that it is unfair, even it is fair, because it is unfamiliar.

As I said, a knockout tournament with a loser's bracket also let's other teams player more games and has fairer matches near the end. As number of participants increase, this only becomes more true. And, as I said before, if a round robin takes too long, and you want exciting matches near the end, do a limited round robin "regular season" to seed an elimination bracket. Then everyone gets to play a set minimum number of games, and the best of the best enter a very short very exciting playoff.

lackofcheese

Alright, that makes some sense now.
This is exactly like the Swiss System; the only difference being that after having sorted by number of wins / losses, you're also sorting by how severe the wins / losses were.

However, Scott makes a very good point in saying this:-

Even so, having better ratings will change who your opponent will be in future rounds. Teams will be encouraged to either run up the score, or crank up the defense, in order to get weaker opponents in future rounds. Rank should only be determined by number of wins and losses. The in-game strategies need to be completely untouched by tournament structure. While playing a game, having more points than the opponent should be the only thing on the mind of the participants. Even a small effect on that tarnishes the purity factor.

Luke Burrage

The Swiss System takes just as many rounds as your system.
You really haven't made a clear case for
1) What's wrong with the Swiss System
2) What makes your system better than it.

1. Nothing is wrong with the Swiss System. It probably works pretty much perfectly for the kind of game and tournament for which it is designed.
2. From the wikipedia page:

The basic rule is that players with the same score are ranked according to rating. Then the top half is paired with the bottom half. For instance, if there are eight players in a score group, number 1 is paired with number 5, number 2 is paired with number 6 and so on.

In chess tournaments the players probably have rankings but what happens if you are in a situation where the players or teams don't have previous rankings? By using score difference as a rough guide to how good a team is compared to their opponents, the system I've outlined here can make closer matches more quickly in the tournament that the Swiss System.

But as you've had a chance to experiment with the spreadsheet you've probably worked that out by now yourself.

Luke Burrage

Alright, that makes some sense now.
This is exactly like the Swiss System; the only difference being that after having sorted by number of wins / losses, you're also sorting by how severe the wins / losses were.

However, Scott makes a very good point in saying this:-

Even so, having better ratings will change who your opponent will be in future rounds. Teams will be encouraged to either run up the score, or crank up the defense, in order to get weaker opponents in future rounds. Rank should only be determined by number of wins and losses. The in-game strategies need to be completely untouched by tournament structure. While playing a game, having more points than the opponent should be the only thing on the mind of the participants. Even a small effect on that tarnishes the purity factor.

This is inherent in another tournament formats you are promoting over this one:

And, as I said before, if a round robin takes too long, and you want exciting matches near the end, do a limited round robin "regular season" to seed an elimination bracket.

Four teams in a round robin group. One team has won a game and drawn a game. In the final game the know they will either go through and face a certain team with a win, go through and face a different team with a draw, or lose and not make it through at all. How is this different? The off-field rules have an effect on the play if the team thinks the "draw" option is better.

a knockout tournament with a loser's bracket

Maybe you are up against a team in the next round who you have been beaten by 100% in previous meetings. But the team you are playing against, you know you can beat, and you also know they can beat the next team too. You look at the losers bracket and think you can beat all those teams easily. Are you going to play to win, or will you let your current opponents take out the harder team in the next round and then take them out in the final? Again, tournament rules influencing play on the field. You can't have it both ways.

And, as I said before, if a round robin takes too long, and you want exciting matches near the end, do a limited round robin "regular season" to seed an elimination bracket. Then everyone gets to play a set minimum number of games, and the best of the best enter a very short very exciting playoff.

The stated aim of this tournament is so everyone gets equal number of games. I'm not just pulling that out of thin air, it was something my friends and I considered before making it.

It's weird in that it is not something the majority of people are familiar with. Even if it did produce a better tournament, the trouble it would cause from a PR perspective would make it difficult. The participants in the tournament are likely to complain that it is unfair, even it is fair, because it is unfamiliar.

For informal tournaments at conventions, I'm not sure we need to worry about PR. Also, when we tried the Swiss System, the majority of the people were unfamiliar with it and didn't understand it fully.

Confusion over my typo cleared up, and knee-jerk reactions over with, I'll end by quoting myself in an attempt to get some other feedback.

I hope you find it interesting or useful or at least give some helpful feedback. I will not respond to anyone moaning about how limited it is. I KNOW that, but this has come in handy a few times to make organizing simple, fair tournaments, and I just want to make it better.

lackofcheese

Well, your system doesn't take into account the fact that beating a strong (higher rated?) opponent should be worth more than beating a weak opponent with the same margin.

Apreche

For informal tournaments at conventions, I'm not sure we need to worry about PR.

That's actually where you have to worry about it the most. Since it is such an informal situation, it is very difficult to sit everyone down and explain something to them. If it was a very formal situation, you could have a little conference in a classroom setting where you explain it to everyone. The confusion at the convention will probably result in decreased fun and participation, which are more important than fairness at a convention.

Luke Burrage

Well, your system doesn't take into account the fact that beating a strong opponent should be worth more than beating a weak opponent.

I know. Do you know a way to work this into the system?

Do you propose that by beating a higher ranked team, or losing to a lower ranked team, modifies the value of the difference between the two scores? Should the points difference count the most and always trump the level of the other team? Eg. if the points difference is 4 in two matches in the second or third round, the difference in the ratings should determine which winning team should come out ahead.

Or should beating a higher team count more even if point difference is unequal? Eg. should beating the (current) 3rd place team by 10 points have more value than beating the bottom placed team by 5 points?

It would make the spreadsheet far more complex, I'm not sure if I'd have to start again from scratch or not. This spreadsheet calculates all of the scores all the time and you only sort the data when a round has finished to line up the next matches. To calculate the "beating a high ranking team" value I'd have to make a real effort somewhere in the guts of the system. I'll give it a go and get back to you.

Luke Burrage

For informal tournaments at conventions, I'm not sure we need to worry about PR.

That's actually where you have to worry about it the most. Since it is such an informal situation, it is very difficult to sit everyone down and explain something to them. If it was a very formal situation, you could have a little conference in a classroom setting where you explain it to everyone. The confusion at the convention will probably result in decreased fun and participation, which are more important than fairness at a convention.

Good points, but I was referring to juggling conventions and games of volleyclub and team combat. Jugglers are usually a bit more laid back and not overly concerned about picky organisation. They leave that up to the organizers, which are often my friends or myself, and an easy spreadsheet which WE know is fair makes life a lot less stressful and therefor more enjoyable for us.

So far there was only one tournament where people were concerned about the system and my friend just said "It uses wins and losses to determine who you play in the next round, and for equal wins and losses, score difference is used as a tie break." Everyone seemed happy enough with that.

For board game conventions and tournaments I'm sure a 5 minute session at the start to explain it in more detail would be needed, but that should probably be done for all systems, as there will probably be at least one person who says "What does double elimination mean?"

lackofcheese

Do you propose that by beating a higher ranked team, or losing to a lower ranked team, modifies the value of the difference between the two scores? Should the points difference count the most and always trump the level of the other team? Eg. if the points difference is 4 in two matches in the second or third round, the difference in the ratings should determine which winning team should come out ahead.

Or should beating a higher team count more even if point difference is unequal? Eg. should beating the (current) 3rd place team by 10 points have more value than beating the bottom placed team by 5 points?

It would have to be a balance between the two, I would think.

Essentially, I think you would use the difference in overall ranking as some kind of scaling factor on the score difference.

HMTKSteve

This sounds a lot like the modified Swiss format used in Decipher Star Wars CCG tournaments (back when the game was alive).

I got royally screwed in one tournament because the first kid I played against sucked so bad I beat him with a 52 point differential (60 cards in the deck). After that I faced all of the toughest players and always won or lost with a 1 to 3 point differential. If I had let one of them cream me it would have knocked my differential down far enough that I would have been playing against easier opponents.

We addressed this problem locally (I used to be a tournament director) by doing our initial pairings based on the players current official ranking. Some players did not like that they were always facing the same opponents but the games ended up being a lot closer in the first round. The second round was often a free win for the top players who lost their first game as they would be up against someone from the lower tier who played well. Those who won continued to battle against others who won and the competition would quickly grow very stiff.

Luke Burrage

I got royally screwed in one tournament because the first kid I played against sucked so bad I beat him with a 52 point differential (60 cards in the deck). After that I faced all of the toughest players and always won or lost with a 1 to 3 point differential.

This kind of thing is a good argument for weighting the differential by the rating of the player. However, in the first round you have no way to tell who is highly rated and who is low rated. So you can only do this from the second round onwards.

Not every game has a ranking or seeding or rating system that carries over from one tournament to the next. I remember being in one tournament where the orgs asked everyone to rate themselves out of a hundred on how good they were. This didn't work out so well as the team I was on rated ourselves 80 which we guessed meant "out of the 24 teams we are in the top 3 or 4" and were put in as the top seed in a group. However, also in our group was a team who had rated themselves 70 for "above average" and another team who rated themselves 50 as in "We think we have a 50% chance at winning the tournament". Those teams knocked us out in the group stage and went on to meet each other in the final. That tournament sucked!

Starfox

Using "seeding" and group stages was annoying because the good teams (our teams) would have to play through a load of easy games before doing anything interesting.

Here's one of your problems. In tournaments for sports and such, you want the two best teams playing in the final game. Seeding teams gives the best teams the best shot at reaching the final. Also, the higher seeded teams get rewarded for doing well in the regular season by not having to play another top-tier team right off the bat. Doing anything "interesting" early on is moot. The only objective is to win.

Luke Burrage

Using "seeding" and group stages was annoying because the good teams (our teams) would have to play through a load of easy games before doing anything interesting.

Here's one of your problems. In tournaments for sports and such, you want the two best teams playing in the final game. Seeding teams gives the best teams the best shot at reaching the final. Also, the higher seeded teams get rewarded for doing well in the regular season by not having to play another top-tier team right off the bat. Doing anything "interesting" early on is moot. The only objective is to win.

I don't think this is my problem, just a problem with any tournament. In the tournament in the spreadsheet, the only way to win is to win all your games, and the only objective you have is to win all your games, and the two teams in the final game are the only two teams not to have yet lost, and the winner will have played against two out of three of the other top half teams.

I know what seeding is for and I understand why it is a good thing to have. However, if you don't have any seeding, this tournament system at least takes a step in the right direction. With 8 teams and 4 rounds, the final game would be between the top two teams, but the chances are good they would already have played each other in the tournament before. Maybe that could be the best thing, actually: if, after the final rankings are shown, the top two teams have not yet played, a final should be held to decide. If the top rated team wins they've gone WWWW and have unrivaled bragging rights. If they lose the two teams will be WWWL and something like LWWW and then... well... then it comes down to the second ranked team winning by enough points to overcome their first loss. Not perfect.

Thanks for your comments.

HMTKSteve

Seeding is designed to keep the audience watching.

Luke Burrage

So over the summer I used this spreadsheet a few times and it always worked pretty well. I can think of a few changes where it can be improved but I need to test them out carefully first.

At one event, the EJC Celebrity Fight Night, the girl who was organising the tournament asked if she could borrow my laptop and this spreadsheet to work out the fixtures. She wanted to run a double elimination knockout with 9 people. I said it wouldn't work, but she used the spreadsheet anyway. Turns out it was a lot easier than keeping track on paper, like she has done for the past few years. This is a 16 player, 4 round version she used:

Of course with 9 players there was always a chance it wasn't going to work out perfectly, and after 4 rounds there were 3 people with three wins and one loss. Thankfully it is possible to have a 3 way match. Thankfully I was the last one eliminated before the final so I could watch, with the other 1000 or so spectators, the very entertaining epic 20 minute battle .