Swiss Tournament Automatic Generator

[Chexhero]

I will add a web service to enable a lookup of a player and their rating, which we can hopefully hook into your tournament generator. Or make the ratings available via a download which we can merge into the program. I'm pretty sure Eric uses ELO so we potentially could show "unofficial" rating changes real-time..... but perhaps I'm getting ahead of myself. I just see great possibilities.

Love the idea here. Not sure if this is what Jason is referring to but perhaps when doing Swiss pairings, players can be matched up in 1st round based on their rating. Players skill level is not really taking into account at all when doing 1st round pairings and so we end up with very lopsided 1st round matchups, giving unfair advantages to some of the better players.

Very good job Vic!

[liam stephens]

Nope ! - The two highest rated players should not be drawn against each other in the first round. The purpose of the Swiss System is (like a bottle of fresh milk) to let the cream gradually rise to the top.

The most common method adopted is to list the players in rating order, then split the list into two halves. (top half and bottom half)
Example - say 50 players.
List them in rating order 1 to 50.
Top half 1 to 25.
Bottom half 26 to 50.
Then in first round 1 plays 26, 2 plays 27 etc., all the way down to 25 plays 50.
Thereafter normal rule applies of pairing players on equal points. (or nearest)

[Alex_Moiseyev]

Then in first round 1 plays 26, 2 plays 27 etc., all the way down to 25 plays 50.

Or 1-50, 2-49, ... 25-26. I saw this kind of pairing in some events. In this case cream will be maximum

[Chexhero]

Nope ! - The two highest rated players should not be drawn against each other in the first round. The purpose of the Swiss System is (like a bottle of fresh milk) to let the cream gradually rise to the top.

The most common method adopted is to list the players in rating order, then split the list into two halves. (top half and bottom half)
Example - say 50 players.
List them in rating order 1 to 50.
Top half 1 to 25.
Bottom half 26 to 50.
Then in first round 1 plays 26, 2 plays 27 etc., all the way down to 25 plays 50.
Thereafter normal rule applies of pairing players on equal points. (or nearest)

Well I am not an expert on how the Swiss system actually works. The way you mentioned, might be better. I am just saying, I rate tournaments all the time where the Swiss pairing is used. And when I look at the matchups, say between the top 3 finishers, I see that the sometimes there is one player who seems to draw a much easier players throughout. Using the ratings to help assist with the pairings might help make a fairer tournament.

[Michael Holmes]

Tennis tournaments do the split. But chess tournaments have three ways to play including highest versus lowest to start first round and then highest vs highest thereafter.

[vhabgood]

Very nice Vic!
Releasing under GPL, are you planning on hosting on google code or anything like that? What are you using to store the information when the program closes? Flat file, xml, or maybe sqlite db?

After a couple of e-mails I received this week it looks like I might be able to add a feature to help with initial pairings.
It looks like from the hard work of JR and John I might FINALLY have access to a full ACF roster.
And with the work of Joe and Eric we now have ratings regularly updated again.

I will add a web service to enable a lookup of a player and their rating, which we can hopefully hook into your tournament generator. Or make the ratings available via a download which we can merge into the program. I'm pretty sure Eric uses ELO so we potentially could show "unofficial" rating changes real-time..... but perhaps I'm getting ahead of myself. I just see great possibilities.

As of now, I have given the user the option to write the results to a text file to be stored in the same directory that the executable/binary resides in. This could easily be adapted to a database if we want to go that direction. I have to fight the urge to run away this idea as well. I did leave room for things to be upgraded in the future though. We will just have to see what everyone wants. I think google code might not be a bad idea. I was thinking originally of just posting it in github, but I am not so sure how well this community is adapted at that sort of stuff.
John and Travis, I will send you guys a message today and you guys can tear it up however you want .
I love these ideas and I will continue to check this forum often to see if there are more ways to improve it.

Merry Christmas!
-Vic

[vhabgood]

Nope ! - The two highest rated players should not be drawn against each other in the first round. The purpose of the Swiss System is (like a bottle of fresh milk) to let the cream gradually rise to the top.

The most common method adopted is to list the players in rating order, then split the list into two halves. (top half and bottom half)
Example - say 50 players.
List them in rating order 1 to 50.
Top half 1 to 25.
Bottom half 26 to 50.
Then in first round 1 plays 26, 2 plays 27 etc., all the way down to 25 plays 50.
Thereafter normal rule applies of pairing players on equal points. (or nearest)

That makes sense to me. Is this the way everyone does it or wants it? If so, I can easily adapt the round generator to behave like this in the first round. Thanks!

-vic

[Alex_Moiseyev]

1-50, 2-49 is more fair and better rank players.

[liam stephens]

Might work in a strong field if the rating band width was narrow, but in most of our tournaments the rating band width is wide and often players can enter whichever section they like, so would only result in the cream soon becoming curdled. Pairing the master players against the weakest opponents would most likely simply gift them (the masters) 4 points - not the best option IMHO. We saw this only too clearly in the last QT, in many of the positions Alex showed on the Forum.

[Alex_Moiseyev]

If the system doesn't work evenly for all groups of players in all cases, random pairing will be best.

In unfortunate reality we never have 50 players, usually 10-14, maximum 20. If we have just 2-3 players out of class and 4-6 relatively even strong players, it is important who plays with weak players in the eginning. Not having the best rating will work better if total number of players is not very big. Top 2-3 players with best rating will be pounished.

The system 1-25, 2-26 etc will be OK if our goal is to punish players with high rating. In this case we have to adjust slightly our ACF/WCDF Constitutions and BY-LAWS

Clear ?

Pairing the master players against the weakest opponents would most likely simply gift them (the masters) 4 points - not the best option IMHO.

Why letting best players get more points is not best option ? Isn't main goal of any system - proper ranking ?

[liam stephens]

Yes, with proper ranking, sharks and minnows would not be allowed to swim in the same pond.

[Alex_Moiseyev]

What if we did groupings within rating classes? Let's say we had a one-division tournament with 12 players: four rated 2000+, four rated 1700-1999, and four rated below 1699. In that case, the first round pairings would be 1v4, 2v3, 5v8, 6v7, 9v12, and 10v11. Now, there may still be some substantial rating disparities at play, but in my opinion it would be a better use of ratings than an absolute sharks-and-minnows pairing.

John, this is much better if rating is accurate I support this idea.

[waynegober]

Never delighted more in a tournament than when the minnow grows teeth and take bite out of one of the sharks......

I always love to see their big smile

So for the fun of it all.... Swim and let Swim

[waynegober]

Over 20 years ago, I was paired with Gene Lindsay(how I got enough points I don't recall) about mid-tournament. I was rated much lower and no where near his level at the time. I was suppose to be Shark snack. Well, I won a game. To this day, I can recall the game and delight in that win....

Give the minnows a chance !!

[Eric Strange]

This is WONDERFUL!!!

I am so glad someone is bringing the ACF forward into the technology age and even more pleased with the acceptance of an automated pairings system within the ACF community.

I Had been planning on getting a web based version of this for a long time now, and just haven't had time with redesigning my site and working with coder to get this applet finished.

Thank you so much for this Vic! If you need anything, just let me know and I am on board 100%

-Eric Strange