Tournament Formats Ideas

[kal5056]

I am organizing a tournament and want to maximize turnout. Here is my issue.
We have a group of player traveling a good distance and they realy want to play 5 rounds (1 Friday Night, 3 saturday, and 1 sunday) to make the most for thier travel. There is another group that would perfer to wait and travel in on Saturday morning and just play 4 rounds.

I am not thrilled with a format whereby you play 5 games if you can and drop your low score to match those that only play 4 games.

I am leaning toward giving players that play 4 games their average score for the 5th score but anm curious how to do pairings from round to round when you have half the players with 2 games and half with 1.

Would you think it best to do pairing based on average score every round? Or does someone have a better method?

Thank You
Gino
SMAC

[Phaze_of_the_Moon]

Pair them as having tied the first round.

[petedalby]

We have a 5 game comp in the UK in April - the BHGS Challenge.

3 games on the Saturday, 2 on the Sunday. For those who arrive on the Friday there is the option of practice games. Their weblink below:

http://www.bhgs.co.uk/

Pete

[DaiSho]

I am organizing a tournament and want to maximize turnout. Here is my issue.
We have a group of player traveling a good distance and they realy want to play 5 rounds (1 Friday Night, 3 saturday, and 1 sunday) to make the most for thier travel. There is another group that would perfer to wait and travel in on Saturday morning and just play 4 rounds.

I am not thrilled with a format whereby you play 5 games if you can and drop your low score to match those that only play 4 games.

I am leaning toward giving players that play 4 games their average score for the 5th score but anm curious how to do pairings from round to round when you have half the players with 2 games and half with 1.

Would you think it best to do pairing based on average score every round? Or does someone have a better method?

Thank You
Gino
SMAC

I'd give them 'bye' points, in other words 10 points each (assuming you're using the 0-25 method) and work them out that way.

Ian

[shall]

I have found the best apporoach to byes si to give 10 points for the sake of deciding the draw and then to swap this for the average score from other games at the end for the result. This seems fair and just. Someone who gets a bye having worked to get an average of 15 points gets 15 points, a player who is a novice and scores 4 on average gets 4 for their bye.

Hope that helps

Si

[sagji]

I think you have to use average score, both for pairing and for final result. Anything else gives one group or the other an advantage.
You might also allow players to enter for FRI+SAT, and even SAT only, but exclude those with less than 4 games from prizes.

I would be inclined to do the daw for game 2 (SAT AM) with the Friday players playing each other based on score, and the Saturday starters playing each other (random draw.) And possibly have a small prize for the Friday starters after the SAT AM game.

[nikgaukroger]

We have a 5 game comp in the UK in April - the BHGS Challenge.

3 games on the Saturday, 2 on the Sunday. For those who arrive on the Friday there is the option of practice games. Their weblink below:

http://www.bhgs.co.uk/

Pete

US comps often seem to only have 1 game on the Sunday - I have assumed this is because of the distances home that can be involved coupled with their miserly holiday allowances. I've seen Challenge type timetable suggested before for US comps and been rejected.

[babyshark]

We have a 5 game comp in the UK in April - the BHGS Challenge.

3 games on the Saturday, 2 on the Sunday. For those who arrive on the Friday there is the option of practice games. Their weblink below:

http://www.bhgs.co.uk/

Pete

US comps often seem to only have 1 game on the Sunday - I have assumed this is because of the distances home that can be involved coupled with their miserly holiday allowances. I've seen Challenge type timetable suggested before for US comps and been rejected.

You hit the nail on the head, Nik. It is a great pity but there it is.

Marc

[shall]

I think you have to use average score, both for pairing and for final result. Anything else gives one group or the other an advantage.
You might also allow players to enter for FRI+SAT, and even SAT only, but exclude those with less than 4 games from prizes.

I would be inclined to do the daw for game 2 (SAT AM) with the Friday players playing each other based on score, and the Saturday starters playing each other (random draw.) And possibly have a small prize for the Friday starters after the SAT AM game.

I looked at that a lot and the porblem is that you don't have a sensible average score to use base on results so far. First and second round results are not great predictors of the average. If you use a real result for a bye with an unrealiable average it causes more problems than its worth and gives groups even more advantage than using 10 in my analysis of it. Using the tourney average pulls the randomness back into line and makes it more even. Nobody high up gets a bye after round 1 anyway. PS how do you pair the guy who had a first round bye if you use the actual average score ... something he doesn't have ??

Si

[lawrenceg]

I looked at that a lot and the porblem is that you don't have a sensible average score to use base on results so far. First and second round results are not great predictors of the average. If you use a real result for a bye with an unrealiable average it causes more problems than its worth and gives groups even more advantage than using 10 in my analysis of it. Using the tourney average pulls the randomness back into line and makes it more even. Nobody high up gets a bye after round 1 anyway. PS how do you pair the guy who had a first round bye if you use the actual average score ... something he doesn't have ??

Si

You seem to be saying that 10 is a better estimate of final average score than using the actual average after 1 or 2 games as the estimate.

Of course, final average score is not the thing you are trying to estimate, which is expected first round score. Lacking any other information, the average of all the other first round scores seems the place to start. That would give a proportion of the 5 bonus points for a win as well as the flat draw 10, depending on how decisive the early games were.

If you were using Glicko ratings you could simulate the bye players in the first round and then slot them in at the appropriate point in the draw. That ought to give a better estimate of what they would have scored in the first round than simply giving them all 10, which would give the better players a slightly easy 2nd round opponent.

I agree with using final average for final placings. It is reasonably fair and easy to understand. It would require a lot of statistical work to come up with a fairer way of doing it, which most people probably wouldn't have confidence in anyway.

[hazelbark]

The other option is to arrange an event of different nature Friday night. Including potentially different lists.

Say a doubles game with 1200 points per side 600 from each list.

[kal5056]

Right now I am leaning toward using a 10 point BYE score for the pairings through round 4.
I would then use the 3 game average from saturday to calculate the BYE score (average) and add that average in BEFORE the final round pairings are decided.

This average would not change and eveyone knows where they stand and need going into the final game.

I want to avoid having to calculate an average including the final game and then making adjustments to totals by anything other than the final round score.

Would hate to see someone think they had won after the final game and then because player #2's average goes up they edge them out when the two did not play head to head on the final table.

I am still digesting this format and appreciate the brainstorming you all are helping me on here.

Gino
SMAC

[kal5056]

I new wrinkle has been proposed. I would appreciate feedback.

Since the players that cannot play on Friday night would be getting the average of thier 3 games on Saturday as thier 4th score prior to going into the final round shouldn't I offer the players that do play the Friday night game the option to use the average of thier 3 games on Saturday if it is higher than thier score on Friday Night?

This provides an incentive to play on Friday night. Does it not?

Can someone point out the possible pitfall to this idea.

I just hate to see someone that gets 25-0'd on Friday night stuck with that 0 when someone that does not play that round will get thier average of Saturday's scores.

Many Thanks
Gino

[Phaze_of_the_Moon]

I new wrinkle has been proposed. I would appreciate feedback.

Since the players that cannot play on Friday night would be getting the average of thier 3 games on Saturday as thier 4th score prior to going into the final round shouldn't I offer the players that do play the Friday night game the option to use the average of thier 3 games on Saturday if it is higher than thier score on Friday Night?

This provides an incentive to play on Friday night. Does it not?

Can someone point out the possible pitfall to this idea.

I just hate to see someone that gets 25-0'd on Friday night stuck with that 0 when someone that does not play that round will get thier average of Saturday's scores.

Many Thanks
Gino

That gives those who play Friday a huge advantage: if they score 25 they keep it, if they score 0 they get three easy pairings then replace it with their average. That is how they say a "free roll".

[sagji]

Right now I am leaning toward using a 10 point BYE score for the pairings through round 4.
I would then use the 3 game average from saturday to calculate the BYE score (average) and add that average in BEFORE the final round pairings are decided.

This average would not change and eveyone knows where they stand and need going into the final game.

I want to avoid having to calculate an average including the final game and then making adjustments to totals by anything other than the final round score.

Would hate to see someone think they had won after the final game and then because player #2's average goes up they edge them out when the two did not play head to head on the final table.

I am still digesting this format and appreciate the brainstorming you all are helping me on here.

Gino
SMAC

The problem with a fixed value is it gives easy games to good players, and hard games to bad players that skip the Friday game. When doing the draw for the last Sat game players that played on Fri range from 0 to 75, those that skipped it range from 10 to 60. For the mid Sat game the ranges are 0-50, and 10-35.

With an average the players that skipped a game will have a greater varience, but it won't be as extreme as the difference for the fixed value. If concerned you could keep the two streams seperate untill after the mid Sat game.

It might be helpfull to have an indication of numbers, and split - the smaller the numbers, and the more uneven the split, the more effect the differences will have.

[shall]

You seem to be saying that 10 is a better estimate of final average score than using the actual average after 1 or 2 games as the estimate.

Not better, but not much worse when you look at it in reality. So in my view not worth the effort.

Of course, final average score is not the thing you are trying to estimate, which is expected first round score. Lacking any other information, the average of all the other first round scores seems the place to start. That would give a proportion of the 5 bonus points for a win as well as the flat draw 10, depending on how decisive the early games were.

If using the +5 system that would be fine by me. I was assuming a simple 0-20.

If you were using Glicko ratings you could simulate the bye players in the first round and then slot them in at the appropriate point in the draw. That ought to give a better estimate of what they would have scored in the first round than simply giving them all 10, which would give the better players a slightly easy 2nd round opponent.

Absolutely. Each player given his expected score from masses of past experience.....feel free to try

I agree with using final average for final placings. It is reasonably fair and easy to understand. It would require a lot of statistical work to come up with a fairer way of doing it, which most people probably wouldn't have confidence in anyway.

Basically my view when I merge the stats and reality. Good enough and vastly better than the common approach of giving a max to someone with a bye. 90/10 rule satisfied IMHO.

Si