PDA

View Full Version : Box Hill Swiss Pairing System



arosar
07-04-2005, 11:00 AM
All this talk about quarantining. What's this Box Hill system all about it anyways? How does it work exactly?

AR

Bill Gletsos
07-04-2005, 01:35 PM
All this talk about quarantining. What's this Box Hill system all about it anyways? How does it work exactly?I cannot be bothered checking but I believe starter has in the past discussed the Box Hill system at length over in his competitive index thread or perhaps it was on the old ACF board.

Garvinator
07-04-2005, 02:14 PM
From my understanding the Box Hill system works as follows:

All players are entered as one complete field, every one in the draw are divided into different quartiles(commonly known as rating groups, except that even the top seed is part of the first group).

Each rating group is a given a +2 point bonus at the start of the tournament. Therefore those who are in the bottom group start on 0 points, the second bottom rating group starting on 2 points and so on and so forth. If there were five rating groups, the top group of players would start on 8 points i think.

Players are then paired as normal including the bonus points. This means that in the first round, players from the same rating groups would be playing each other which reduces the rating differential of each pairing from a normal swiss first round.

The bonus points are added to each persons score for the entire tournament.

You would hear the term intermingling. This means that after a few rounds, some of the players in say the C quartile have scored enough points that they are now even with the players in the B quartile including all the bonus points having being added. This means that the B and C players are playing each other, so they are intermingling.

In the Box Hill system, only players from the A quartile can win outright first place prizes, the trade off being that everyone gets more games close to their own rating.

For players that are playing well and out performing their rating, then later in the tournament they will get opponents who are stronger than them.

Recherché
07-04-2005, 05:02 PM
All this talk about quarantining. What's this Box Hill system all about it anyways? How does it work exactly?

AR

Garvin described it reasonably well. Basically it enables a very large tournament (such that most of our members can play in it) to be run much more effectively (and over fewer rounds, perhaps) than a monster swiss of the same size.

Players in each divisions are competing for the prizes in their division, regardless of whether or not they end up having games against people in other divisions, above or below. Some (non Box Hill, I'd like to emphasise) people have criticized the fact that B division players can't win A-division prizes, but this system is actually much better for those players with regards to their chances of winning prizes. I can attest to this myself, having won several divisional firsts.

Instead of weaker players getting paired immediately against the strong players in the field as in a normal swiss, players who perform strongly in their division earn the right to play the players in the division above. Players who do badly in their division get the chance to play easier opponents from the division below.

It all works very well. The only problem as I see it comes when the finishing places (and hence prizes) are to be decided in the final round. In some cases the divisions have been quarantined (ie. no intermingling, B-division plays only B-division), in others cases they have not. I am in favour of quarantining, for the reasons outlined earlier in this thread.

Garvinator
07-04-2005, 05:14 PM
It all works very well. The only problem as I see it comes when the finishing places (and hence prizes) are to be decided in the final round. In some cases the divisions have been quarantined (ie. no intermingling, B-division plays only B-division), in others cases they have not. I am in favour of quarantining, for the reasons outlined earlier in this thread.
and some need more convincing ;) one question i do have and i think i have asked this before, lets say by rating a particular player is due to be in the c quartile, can they ask to begin the tournament in the a or b quartile?

pax
07-04-2005, 05:24 PM
I can't really see any particular advantages over having separate divisions in the usual way (either Swiss or round robin). The difficulty is that in later rounds the top B division players may be playing bottom A division players, rather than playing each other to determine a winner in that division (i.e there will be fewer games between the top players in each division except A).

There are certainly advantages over a monster swiss (especially if there aren't many rounds).

ursogr8
07-04-2005, 05:37 PM
I can't really see any particular advantages over having separate divisions in the usual way (either Swiss or round robin). The difficulty is that in later rounds the top B division players may be playing bottom A division players, rather than playing each other to determine a winner in that division (i.e there will be fewer games between the top players in each division except A).

There are certainly advantages over a monster swiss (especially if there aren't many rounds).

hi pax

The reason you can't see the advantages is that Garvin and Rob have simply concentrated on the mechanics as requested by Amiel in his post.
What their two replies lack is the reason for each decision that leads to designing this tournament format.
I will supply later today (time permitting).
Your other point about B's not playing each other in later rounds ....is simply Rob's question repeated. I have not commented on his point, but will do so.

starter


and some need more convincing ;) one question i do have and i think i have asked this before, lets say by rating a particular player is due to be in the c quartile, can they ask to begin the tournament in the a or b quartile?
No. It defeats the whole purpose of the exercise. But let me begin at the start, bbl.
starter

Recherché
07-04-2005, 05:53 PM
and some need more convincing ;) one question i do have and i think i have asked this before, lets say by rating a particular player is due to be in the c quartile, can they ask to begin the tournament in the a or b quartile?

Not generally, no. Though in a "JGB scenario" arbiter/dop discretion might well be employed.


I can't really see any particular advantages over having separate divisions in the usual way (either Swiss or round robin). The difficulty is that in later rounds the top B division players may be playing bottom A division players, rather than playing each other to determine a winner in that division (i.e there will be fewer games between the top players in each division except A).

There are certainly advantages over a monster swiss (especially if there aren't many rounds).

The advantages are considerable:

1) All the players in the club are able to play together. This is good for the ratings pool, good for players to learn and improve, and good for the social life of the club.

2) The results for the lower divisions (what would have been ratings groups in a Monster Swiss) are much more reliable, and impressive tournament performances from lower rated players really stand out in a way they don't in a regular swiss. The quarantining I have been arguing for enhances this even further, in my opinion.

3) The games are significantly more competitive than in a standard swiss. There are no junk rounds at all. See starter's "how competitive do you want it to be?" thread for some stats and analysis relating to this (if you dare).

Note that not only are players with similar ratings playing each other more with this system, but lower rated players doing well soon (from round 3 onwards) meet higher rated players doing badly, so there is matching of standards within the current tournament as well as ratings.

4) As an extension to point 3, weaker players are not saddled with near impossible first round games. Moreover, lower rated players earn the right to play higher rated players through doing well in the tournament - a concept I think has significant merit.

5) A tournament like this is much easier to organise and pair than a tournament with two, three or more separate divisions (which is essentially like running two or more separate tournaments).

Recherché
07-04-2005, 05:55 PM
The difficulty is that in later rounds the top B division players may be playing bottom A division players, rather than playing each other to determine a winner in that division (i.e there will be fewer games between the top players in each division except A).

This is actually an advantage in the middle of the tournament. It happens from about round 3, and Box Hill tournaments are typically 7, 8, or occasionally 9 rounds. I only think it's a problem in the last round.

pax
08-04-2005, 09:29 AM
The advantages are considerable:


You didn't read my post. I said I can't see advantages over a tournament split into divisions in the usual way. Your advantages are mostly advantages over a monster swiss.

Re point 5, I would point out that pairing round robins is a piece of cake (you can pair the whole tournament from day 1). Pairing a three division swiss is barely more trouble than pairing a big monster swiss.

The point about the whole club "playing together" is pretty meaningless. Three divisions can play in the same room just as a monster swiss, so socially they are equal. Does the fact that I have a theoretical chance of playing someone from a higher division have any social advantages? I don't think so.




1) All the players in the club are able to play together. This is good for the ratings pool, good for players to learn and improve, and good for the social life of the club.

2) The results for the lower divisions (what would have been ratings groups in a Monster Swiss) are much more reliable, and impressive tournament performances from lower rated players really stand out in a way they don't in a regular swiss. The quarantining I have been arguing for enhances this even further, in my opinion.

3) The games are significantly more competitive than in a standard swiss. There are no junk rounds at all. See starter's "how competitive do you want it to be?" thread for some stats and analysis relating to this (if you dare).

Note that not only are players with similar ratings playing each other more with this system, but lower rated players doing well soon (from round 3 onwards) meet higher rated players doing badly, so there is matching of standards within the current tournament as well as ratings.

4) As an extension to point 3, weaker players are not saddled with near impossible first round games. Moreover, lower rated players earn the right to play higher rated players through doing well in the tournament - a concept I think has significant merit.

5) A tournament like this is much easier to organise and pair than a tournament with two, three or more separate divisions (which is essentially like running two or more separate tournaments).

Might be an idea to split this thread into one specifically about the Box Hill system, as we are no longer discussing the Doeberl.

Recherché
09-04-2005, 03:25 PM
The point about the whole club "playing together" is pretty meaningless. Three divisions can play in the same room just as a monster swiss, so socially they are equal. Does the fact that I have a theoretical chance of playing someone from a higher division have any social advantages? I don't think so.

You're more likely to talk to someone and get to know them if you've played against them. Post-game analysis is a pretty effective ice-breaker.

ursogr8
12-04-2005, 12:50 PM
You didn't read my post. I said I can't see advantages over a tournament split into divisions in the usual way. Your advantages are mostly advantages over a monster swiss.




While the responses of Garvin and Rob concentrate on the mechanics of these possible tourney-designs it will never be apparent what the benefits are; this is because the benefits are derived from player-behavioural influences rather than logical cause and effect.



First, in a usual BHCC field there are 70 or so chess-desparates who will enter any event as long as there is a clock to bash, and a king to knock to the floor when castling. These are salt of the earth people.
But there are others who are more selective about entering tournaments, and are inclined to enter only Grades events, or Interclub events. These folk want competitive games. They swell our fields to 95-120 when we present the right tournament format. This group don’t want the large all-in SWISS.

So, second, we are then led to the concept of running divisions so that players have competitive games. For example, a 100-players field could be split into two tournaments; call them A and B divisions, with say a rating split at 1450. However, as soon as this is announced then substantial lobbying for exceptions (see gg’’ ‘s question, and see the refce to the ‘JGB effect’). I liken it a humorous episode I observed at a seminar morning tea some time back; attendees were asked to mark on a survey piece of paper whether they were above-average drivers or below-average drivers. When the votes were counted, 90% of folk saw themselves as above-average.
Translated to our two division chess environment you can anticipate that many in the category 1300-1450 really see themselves as >1450 players. They want to be in division A. And if they are not, then they are unlikely to enter the tournament. Add to this, the intense pressure from coaches (of juniors) that their charges play UP, not DOWN.
Pax, if you don’t believe any of what I writing here about behaviour….just ask Doeberl (Shaun has already written on this) organisers why they have to artificially allocate $600 to try to keep folk in the MINOR instead of the MAJOR. Even with this golden handcuff there are MINOR players who elected to play UP (out of their rating division) in the MAJOR. For example, two juniors, already exposed in the Doeberl thread, would probably have preferred to walk home with $600 in their kick instead of the alternate ‘look how well I did against some 1800’s.’ But someone persuaded them to play UP. Incidentally, a total of 14 players decided to play UP in the MAJOR (that is players rated U1600 played in the MAJOR even though the MINOR U1600 $600 1st prize, was on offer).
The summary of all this, is that we need to have competitive games to maximise the size of the field and we need to have intermingling of the fields if we are to have divisions as the solution to making the games competitive. We don’t have $600 to spend to encourage some folk to play where we, as organisers, want them to play against their preference.



Thus we come to the third part of the design…how to achieve intermingling?
Our current practice (in a 100-player field) is to permanently accelerate the A division by 2 points. As Rob describes, this produces the first intermingling in round 3. The new issue that arises is that the best B division players can finish 7 rounds of a divisionalised_intermingled_SWISS on 5/7, (whereas in an all-in standard SWISS they stretch to get 3/7, or 3.5/7). The best performed B division players might well get to play 3 or 4 opponents from A division in the last 4 rounds; each opponent will be in the very low end of A division. The B division players need to clearly understand why they are getting through intermingling
>fairly easy opponents in A
>a score like 5/7
>vetoed from winning A division prizes.
We have found this to be an achievable understanding (with the B’s) by using publicity and T&C’s on the entry form. The B’s understand they get * intermingling * and, more reliable results for rating prizes, and only have to forgo the chance of 1st prize outright as a trade-off.

After all this explanation, I am left with only one remaining issue;...quarantining. This is Rob's word for the practice of making the B's play only B opponents in round 7 of a 7-round SWISS. This is an interesting issue, but not impacting the 'benefits' issue that you raised though. So, I will leave it for further later posts with Rob.

In summary, the style of tournament we have designed has the objective of increasing the number of entries.


starter

jase
12-04-2005, 01:46 PM
Thanks for that explanation Starter. It's an interesting idea and great that organisers are willing to break the mould with regard to the tournaments they create.

I'm interested to hear the views of players as to their experiences in these sort of events - you give the impression they like the concept and are voting with their feet. The club must have done an excellent job of explaining the concept, as in my experience chess players as a subgroup are quite resistent to change.

Has the club also previously flirted with accelerated Swiss systems, which seek to achieve a similar function to the pairing system you've outlined? Both seem to have the same objective; both are also quite reliant on accurate ratings to achieve their desired outcomes.

Mischa
12-04-2005, 01:57 PM
Starter, my junior was automatically placed 'up' by the organisers...this is in the Rookies cup.

Davidflude
12-04-2005, 02:05 PM
These are my opinions and I have played many Swiss Tournaments over many
years.

The problem with a normal (or common garden variety) swiss in a large tournament with very wide disparities in playing strength is that players close to the middle of the draw get few games against players of similar strength. Take a look at the ratings of my opponents at this years Begonia. In previous rounds it has been even worse - Grandmaster one round 1300 player the next. (Of course in all other rspects the Begonia was exellent.)

One valid solution is to split the field into multiple tournaments as is done at the Doeberl Cup. This worked very well the two years that I attended.

What we use at Box Hill is another solution which also works well. You get lots of games against opponents where the rating difference is reasonable. If a player plays above his or her rating then the opponents played will become progressively tougher. If a higher graded player is out of form then the opponents become progressively weaker.

It is much more enjoyable to play in a tournament where you get large numbers of pairings against opponents of similar strength rather than playing
players graded 600 points higher one round then 600 points lower the next.

Personally I regard the Box Hill System as being nothing more nor less than
a permanently accelerated pairing system.

ursogr8
12-04-2005, 02:16 PM
Starter, my junior was automatically placed 'up' by the organisers...this is in the Rookies cup.

hi Mischa

Yes, James has probably been UPed in the Rookies Cup, in the past year, because he is amongst the strongest regular juniors that play in this event. But I would be interested to hear if you think he misses something by being UPed in this case? It is not as though there are $600 golden handcuffs in the Rookies Cup that he misses out by being UPed.


But, having said that, you will notice my advert. for next Sunday's Rookies Cup indicates a movement away from prizes based on AGE groups. Instead, we will award rating prizes in about-to-revealed groups. We can do this now that most entrants have fairly reliable RAPID ratings.

regards
starter

ursogr8
12-04-2005, 02:30 PM
Thanks for that explanation Starter. It's an interesting idea and great that organisers are willing to break the mould with regard to the tournaments they create.

:) :cool:


I'm interested to hear the views of players as to their experiences in these sort of events - you give the impression they like the concept and are voting with their feet. The club must have done an excellent job of explaining the concept, as in my experience chess players as a subgroup are quite resistent to change.
You are quite correct jase, explaining how it works is a key to acceptance. That is where bulletin boards can be very handy. But also, we utilise newsletters etc.



Has the club also previously flirted with accelerated Swiss systems, which seek to achieve a similar function to the pairing system you've outlined? Both seem to have the same objective; both are also quite reliant on accurate ratings to achieve their desired outcomes.

I think we tried every variant of a accelerated SWISS system we could imagine. The weakness of these variants occurs when (and if) the acceleration is removed........then the junk rounds re-emerge immediately.

The essence of the long post above is that the acceleration is not removed. There are in fact two division, two tournaments, but with intermingling.
But a community consensus is necessary, to make it successful; which you quickly saw.

regards
starter

arosar
12-04-2005, 02:43 PM
Well, FMD! Now I get it. It sounds like an OK system. I reckon we should try it here.

AR

ursogr8
12-04-2005, 02:54 PM
Well, FMD! Now I get it. It sounds like an OK system. I reckon we should try it here.

AR

hi Amiel


Does this mean you retract your remarks on the COMPETITIVE thread? :uhoh: ;)


starter

arosar
12-04-2005, 03:03 PM
What are you bloody talking about? Your CI is still all crap. That's all there is to it.

AR

Mischa
15-04-2005, 12:54 PM
Starter,I don't think James misses out on anything by being 'uppud' quite the contrary.
I just meant that it was not a 'pushy' mother who did it.
If not age groupings, then what? How will it effect the stronger juniors?
James seems only to play the adults....

ursogr8
15-04-2005, 03:33 PM
Starter,I don't think James misses out on anything by being 'uppud' quite the contrary.
Well noidea, it just depends on how you view the chance to win the $600 Minor prize outright. If you discount this chance to nothing then I would agree that being upped did not miss a financial opportunity.


I just meant that it was not a 'pushy' mother who did it.

I must have missed a/c's poll on who had 'pushy' mothers. ;)


If not age groupings, then what? How will it effect the stronger juniors?
James seems only to play the adults....

Can you expand on this bit....not sure what you mean...can be Sunday?


starter

Garvinator
16-04-2005, 01:56 AM
After all this explanation, I am left with only one remaining issue;...quarantining. This is Rob's word for the practice of making the B's play only B opponents in round 7 of a 7-round SWISS. This is an interesting issue, but not impacting the 'benefits' issue that you raised though. So, I will leave it for further later posts with Rob.
Since Rob hasnt replied to this paragraph of yours starter, I will ask Rob a couple of questions.

Rob, I am starting to think quaranting in the final round might have some merit. The reason I have worked out, i think :uhoh: , is that in the final round, normally there would be two players who have progressed up the food chain into the higher quartiles, leaving the rest of their 'rating' group playing against themselves.

Question though, what happens when you get three or more players on the same score in the final round who are in the situation above?

Mischa
16-04-2005, 10:32 AM
Will talk to you tomorrow Starter, I think we have our wires crossed.

Recherché
16-04-2005, 01:36 PM
Since Rob hasnt replied to this paragraph of yours starter, I will ask Rob a couple of questions.
Starter isn't waiting for me to reply to him. I've already outlined my position. :)


Rob, I am starting to think quaranting in the final round might have some merit. The reason I have worked out, i think :uhoh: , is that in the final round, normally there would be two players who have progressed up the food chain into the higher quartiles, leaving the rest of their 'rating' group playing against themselves.

Question though, what happens when you get three or more players on the same score in the final round who are in the situation above?

Typically it's more than two b-division players who will get an a-division pairing in the final round. That is not to say they're all getting opponents of similar strength, however.

In a quarantined final round the pairing proceed as per normal pairing rules. If, at the top of B-division, there are three people on, say, 5.5/8 going into the final round of a 9 round tournament, then probably you would get two of them playing each other, and the other one playing a b-division player on 5/8. Obviously it's going to depend on the specifics of the situation, but generally the top few players in b-division will not have played each other prior to this round, so they do get paired together if you quarantine it.

ursogr8
16-04-2005, 07:44 PM
Starter isn't waiting for me to reply to him. I've already outlined my position. :)

Typically it's more than two b-division players who will get an a-division pairing in the final round.

hi Rob
In fact at the Autumn Cup round 7 (http://www.boxhillchess.org.au/e2005/e0502fac/crosstable.htm)
there are a swag of players at the top of B who got paired with A Division laggards. At a quick look, all who were chances of winning B were paired up into A.


That is not to say they're all getting opponents of similar strength, however.

The theoretical 'players of equal (similar) strength' would be those on the same rating; and that is just not an option available in the field. I challenge you to find 'quarantined pairings that could have given opponents of the same rating.


In a quarantined final round the pairing proceed as per normal pairing rules. <snip>

This is essentially an argument for giving the best-performed B Division players an easy opponent in the last round of a 7 round tournament. If they are quarantined to B Division opponents then, by force, those opponents have lower ratings than (unquarantined) A opponents. I can't see why you would want to make it relatively easy for these well-prformed B's in their last game.


It might be time to point to specifics in the actual pairings (see link) and higlight any you see as unfair.


starter

Garvinator
16-04-2005, 08:10 PM
Also with quarantining I see two potential scenarios:

1) Three players from Division B are tied for first in Div B. What happens then? Are only two players quarantined? If so what happens to the third player, does that third player have to play a Div A player? Or is the third player also quarantined? If the third player is also quarantined, who does that person play? I would take it he would also have to play a Div B player, who is half a point behind. Then who would you choose to be the fourth player?

2) One player is in front of a group of players, say half a point behind. How many players do you quarantine then? Similiar questions arise as to my first scenario.

Recherché
17-04-2005, 06:39 PM
The theoretical 'players of equal (similar) strength' would be those on the same rating; and that is just not an option available in the field. I challenge you to find 'quarantined pairings that could have given opponents of the same rating.

That isn't the goal of the quarantining. I was just making sure Garvin noted that A-division opponents aren't all created equal.


This is essentially an argument for giving the best-performed B Division players an easy opponent in the last round of a 7 round tournament.

No, it isn't. With B-division quarantined, the B-division players are playing other B-division players on the same score as them (or perhaps half a point off). In this situation, a B-division player in first-prize contention (unless they are well ahead of the field) will be paired against someone who has about the same level of tournament performance (according to the tournament structure) as their A-division opponent. For example, a B division player on 5/8 instead of an A-division player on 3/8.

Because of the way swiss pairings work, B-division players with enough points will more or less always be paired against an A-division player rather than a B-division player with the same number of points as them. That's why the quarantining is necessary, because otherwise the top B-division players will never play each other.

My central point is the simple idea that if two players are leading B division, they should be playing each other in the final round to decide first prize, rather than A-division opponents (which can be a bit of a lottery). This also applies to those competing for second or third in B-division.

Their opponents a half or a full point behind them going into the final round will always be getting easier games, be they against A or B division opponents. However another difference quarantining makes is that if the two b-division leaders are playing each other, they can't both be overtaken by someone half a point behind who hasn't faced a field anywhere near as tough. At most the half-point behind player can get a share of the lead if they draw.


It might be time to point to specifics in the actual pairings (see link) and higlight any you see as unfair.

This is something I may have to revisit later; I don't have time for it now.


Also with quarantining I see two potential scenarios:

1) Three players from Division B are tied for first in Div B. What happens then? Are only two players quarantined? If so what happens to the third player, does that third player have to play a Div A player? Or is the third player also quarantined? If the third player is also quarantined, who does that person play? I would take it he would also have to play a Div B player, who is half a point behind. Then who would you choose to be the fourth player?

2) One player is in front of a group of players, say half a point behind. How many players do you quarantine then? Similiar questions arise as to my first scenario.

The entire division is quarantined. Sorry, I thought that had been clear from the beginning.

ursogr8
17-04-2005, 08:01 PM
No, it isn't. With B-division quarantined, the B-division players are playing other B-division players on the same score as them (or perhaps half a point off). In this situation, a B-division player in first-prize contention (unless they are well ahead of the field) will be paired against someone who has about the same level of tournament performance (according to the tournament structure) as their A-division opponent. For example, a B division player on 5/8 instead of an A-division player on 3/8.

Rob
It is eerie that I anticipated that you would turn to measure performance, at an in-progress, as an indicator of the strength of the opposition, instead of the other metric...their long-term rating. I would have thought that rating is a more reliable indicator of strength, rather than the rather flukey performance in the first 6 rounds of a 7 round SWISS.
If you admit that rating is the superior metric, then by force, an opponent who is A Division is theoretically harder to play than even a well-performing B opponent. Hence my comment that quarantining forces relatively easy last round pairings.



Because of the way swiss pairings work, B-division players with enough points will more or less always be paired against an A-division player rather than a B-division player with the same number of points as them. That's why the quarantining is necessary, because otherwise the top B-division players will never play each other.

Again...this generalisation needs specific examples to be examined....will wait on you.



My central point is the simple idea that if two players are leading B division, they should be playing each other in the final round to decide first prize, rather than A-division opponents (which can be a bit of a lottery). This also applies to those competing for second or third in B-division.

Yes, your idea is simple...is clear... I just don't agree with the objective. But first we need to resolve the earlier points of difference above.


Their opponents a half or a full point behind them going into the final round will always be getting easier games, be they against A or B division opponents. However another difference quarantining makes is that if the two b-division leaders are playing each other, they can't both be overtaken by someone half a point behind who hasn't faced a field anywhere near as tough. At most the half-point behind player can get a share of the lead if they draw.



This is something I may have to revisit later; I don't have time for it now.

keep posting.


starter

ursogr8
17-04-2005, 08:08 PM
Will talk to you tomorrow Starter, I think we have our wires crossed.

Yes noidea...I did have a crossed wire on your point.
It was good today to see Solomon intervene in the SWISS pairings and assure absolutely equal opportunity for the two protagonists in the last round. See, ....the gods do listen.

Btw........neither had much chance against Jesse Jaeger, as it transpired. I mean, not even those guns > Flude, Bonning, Forace, Muthusamy...could get their candles lit.

Thanks for the sterling volunteer work.

starter

Recherché
17-04-2005, 10:02 PM
If you admit that rating is the superior metric, then by force, an opponent who is A Division is theoretically harder to play than even a well-performing B opponent. Hence my comment that quarantining forces relatively easy last round pairings.

Hmm, yes, conceded. In general it should mean easier games for B-division players in the final round. I don't see this as a flaw, however. After all, the intermingling divisions system gives B-division players easier games for at least the first two rounds, and arguably throughout the tournament.


Again...this generalisation needs specific examples to be examined....will wait on you.

Rounds 3, 4, 5, 6 and 7 of the 2004 Box Hill Open (http://www.boxhillchess.org.au/e2004/e0409fop/event.htm).

B-division players were always paired against A-division players two points below them whenever such players were available. I'm sure the same would apply to any other Box Hill tournament which uses the intermingling system. It is an inevitable consequence of the way the Swiss pairing system uses ratings to determine pairings.

Note that this effect is a good thing for most of the tournament.


Yes, your idea is simple...is clear... I just don't agree with the objective.

What is wrong with the objective?

Garvinator
17-04-2005, 10:18 PM
Hello Recherche,

I cant open your link for 2004 Box Hill Open, i think that link is a dud?

Garvinator
17-04-2005, 10:20 PM
B-division players were always paired against A-division players two points below them whenever such players were available. I'm sure the same would apply to any other Box Hill tournament which uses the intermingling system. It is an inevitable consequence of the way the Swiss pairing system uses ratings to determine pairings.
I would need to see more information on this point because sp should be pairing players on the same score, not two points below the Div B players in question.

Have you been following the top half v bottom half thread in the arbiters corner?

Recherché
17-04-2005, 11:47 PM
Hello Recherche,

I cant open your link for 2004 Box Hill Open, i think that link is a dud?
Fixed. It had a carriage return in it that I didn't catch, the forum converted it into a <br> tag.


I would need to see more information on this point because sp should be pairing players on the same score, not two points below the Div B players in question.
A-division is permanently accelerated by 2 points. SP sees them as having the same number of points as B-division players two points above them.


Have you been following the top half v bottom half thread in the arbiters corner?
No.

Garvinator
18-04-2005, 12:08 AM
A-division is permanently accelerated by 2 points. SP sees them as having the same number of points as B-division players two points above them. you seem to think i dont know much about the box hill system and how it works. I do. I recommending it be tried for one of our club tournaments up here. I am not going to recommend something that I dont feel I understand well. ;)

Yes the A div players are two points below a Div B player in real terms for the situation you talk about, but that A div player started off by playing other A div players and so would have started off playing stronger players than the Div B player.

ursogr8
18-04-2005, 07:49 AM
Hmm, yes, conceded. In general it should mean easier games for B-division players in the final round. I don't see this as a flaw, however. After all, the intermingling divisions system gives B-division players easier games for at least the first two rounds, and arguably throughout the tournament.

Rob
I don't think you can argue this line at all.
I thought the point at issue was
a) two divisions, intermingling, quarantine round 7
v
b) two divisions, intermingle, no quarantine round 7.
If this is the debating point, then earlier round opportunities are common to a) and b).
The question is,
in round 7 should we
A) quarantine and hence give relatively easy games, but ensure B v B games.
B) no quarantine and allow A v B games




What is wrong with the objective?

Will discuss next Friday.


starter

Recherché
18-04-2005, 09:55 AM
you seem to think i dont know much about the box hill system and how it works. I do.

No, I thought you did. Your comment surprised me. If you understand it, then what on earth did you mean by "sp should be pairing players on the same score, not two points below the Div B players in question"?

The acceleration does not show up on the score table. Although A-division has a two-point bonus, the players don't see it in the scores. Only SP does.


Yes the A div players are two points below a Div B player in real terms for the situation you talk about, but that A div player started off by playing other A div players and so would have started off playing stronger players than the Div B player.

So what's your point? SP doesn't care about the strength of the field the A-player has faced, it cares about pairing players on the same score based on their rating, the colour they're due to have, and so on.

Recherché
18-04-2005, 10:02 AM
Rob
I don't think you can argue this line at all.
I thought the point at issue was
a) two divisions, intermingling, quarantine round 7
v
b) two divisions, intermingle, no quarantine round 7.
If this is the debating point, then earlier round opportunities are common to a) and b).

Yes, of course they are common. My point was that if a feature of the quarantining from (a) was similar to a desirable feature from the intermingling common to both, then it wouldn't be right to automatically class the quarantining as flawed based on that feature.

Easier games aren't necessarily a bad thing. They may also be more competitive games. I used the early rounds as an example of this, and I think it applies to a quarantined last round as well - the B-division players are playing other B-division players with a similar performance over the tournament; you would expect quite competitive games in those circumstances. For instance my game in the final round of the quarantined 2004 club championship against Surya was a very close affair.


The question is,
in round 7 should we
A) quarantine and hence give relatively easy games, but ensure B v B games.
B) no quarantine and allow A v B games

More or less. Although I think you put too much emphasis on the "easy games" aspect, and also that your choice of the word "easy" is loaded.

But yes, that's the question, and I advocated option (a).

ursogr8
18-04-2005, 10:11 AM
More or less. Although I think you put too much emphasis on the "easy games" aspect, and also that your choice of the word "easy" is loaded.
Rob
You are not the first person on this bb to comment/complain that my choice of words is inconvenient for your side of the argument. Bill and K. have expressed the same unease. :uhoh:


But yes, that's the question, and I advocated option (a).

This is a step fwd...we agree on what we are debating.



Easier games aren't necessarily a bad thing.

:eek: :confused: So you can use the word easier, but I can't call them relatively easy? ;)




They may also be more competitive games.


And they may not be.


I used the early rounds as an example of this, and I think it applies to a quarantined last round as well - the B-division players are playing other B-division players with a similar performance over the tournament; you would expect quite competitive games in those circumstances. For instance my game in the final round of the quarantined 2004 club championship against Surya was a very close affair.

I will look at the data during the week.


starter

ursogr8
18-04-2005, 01:18 PM
I will look at the data during the week.

starter

Rob

From the Autumn Cup, by my reckoning, players 47,48,49,50,51,52, and 57 could have had visions of winning after round 6.

Just picking a few
> player 48 had played 5 Bs
> player 49 had played 5 Bs
whereas
>> player 50 had played 2 Bs and 4 As
>> player 52 had played 2 Bs and 4 As
You don't think 50 and 52 would have been a little narky if a 'constraining fiddle' (sorry, those loaded words slip in again,...strike 'constraining fiddle' and replace with quarantining), :uhoh: , had forced B opponents on 48 and 49 yet again?

starter

Recherché
18-04-2005, 03:03 PM
Rob

From the Autumn Cup, by my reckoning, players 47,48,49,50,51,52, and 57 could have had visions of winning after round 6.

Just picking a few
> player 48 had played 5 Bs
> player 49 had played 5 Bs
whereas
>> player 50 had played 2 Bs and 4 As
>> player 52 had played 2 Bs and 4 As
You don't think 50 and 52 would have been a little narky if a 'constraining fiddle' (sorry, those loaded words slip in again,...strike 'constraining fiddle' and replace with quarantining), :uhoh: , had forced B opponents on 48 and 49 yet again?

starter

50 isn't a good example, because he was a point behind the B-division leaders going into the final round. But lets look at 47, 52 and 57, since they were the three leaders going into the final round. We'll also look at 49, the only player who could pass them to outright first.

47 was paired against a 1627 player, 326 points above his rating.
52 was paired against a 1599 player, 360 points above his rating.
57 was paired against a 1680 player, 490 points above her rating.

49 was paired against a 1402 player, 123 points above his rating.

Now, according the the predictions of the ratings, 49 is far more likely to win his game than any of the other three are. And if those three had all lost, and 49 had won, 49 would have leapfrogged all three to take outright first.

If that had happened, 49 would have won the tournament after facing a significantly weaker field than 47, 52, and 57 and also without playing any of those three. Under those circumstances how can his performance in the tournament be said to be superior to theirs? I don't think it can, and that's the unfairness inherent in an unquarantined final round.

Now, lets look at what a quarantined final round might have looked like:

57 paired against 47
49 paired against 52

Suddenly we get more competitive games, for starters. Ratings differentials of 111 and 40, respectively. They should be competitive based on ratings. Also they are on the same number of points, at the head of the B-division field (except for 49 who is half a point behind). So their games are predicted to be competitive according to tournament performance as well. Neither of those games could be considered easy games for any of them.

The second and most important effect is that 49 cannot jump all three to take outright first. If 49 wins his game then he either gets outright second, or =1st with two others (if the top pairing draw), which is a result much more reflective of his performance during the tournament.

As we can see from the results, 57 and 52 managed draws, and 49 lost, so in the end the result did fairly reflect tournament performance. My point is that the tournament structure allowed for the significant possibility of an unfair result. Quarantining addresses this problem, and addresses it well.

ursogr8
18-04-2005, 04:09 PM
50 isn't a good example, because he was a point behind the B-division leaders going into the final round. But lets look at 47, 52 and 57, since they were the three leaders going into the final round. We'll also look at 49, the only player who could pass them to outright first.

47 was paired against a 1627 player, 326 points above his rating.
52 was paired against a 1599 player, 360 points above his rating.
57 was paired against a 1680 player, 490 points above her rating.

49 was paired against a 1402 player, 123 points above his rating.

Now, according the the predictions of the ratings, 49 is far more likely to win his game than any of the other three are. And if those three had all lost, and 49 had won, 49 would have leapfrogged all three to take outright first.

If that had happened, 49 would have won the tournament after facing a significantly weaker field than 47, 52, and 57 and also without playing any of those three. Under those circumstances how can his performance in the tournament be said to be superior to theirs? I don't think it can, and that's the unfairness inherent in an unquarantined final round.

Now, lets look at what a quarantined final round might have looked like:

57 paired against 47
49 paired against 52

Suddenly we get more competitive games, for starters. Ratings differentials of 111 and 40, respectively. They should be competitive based on ratings. Also they are on the same number of points, at the head of the B-division field (except for 49 who is half a point behind). So their games are predicted to be competitive according to tournament performance as well. Neither of those games could be considered easy games for any of them.

The second and most important effect is that 49 cannot jump all three to take outright first. If 49 wins his game then he either gets outright second, or =1st with two others (if the top pairing draw), which is a result much more reflective of his performance during the tournament.

As we can see from the results, 57 and 52 managed draws, and 49 lost, so in the end the result did fairly reflect tournament performance. My point is that the tournament structure allowed for the significant possibility of an unfair result. Quarantining addresses this problem, and addresses it well.

Rob

Such a long post and yet you don't address my question about the disparity in fields met in 6 rounds.

Instead you launch into a 'if my aunty had balls she would be my uncle', discussion.

And yes, the intermingling in round 7 did produce largish rating differentials for the top B Division players. This is what they strive for. THis is what their coaches want.

starter

Recherché
19-04-2005, 10:34 AM
Such a long post and yet you don't address my question about the disparity in fields met in 6 rounds.

My answer was that quarantining the final round doesn't make the field less fair to those at the top - it actually makes it more fair.

As my stats should have made clear, keeping the intermingling not only allows leapfrogging, but the players just behind the leaders in B-division get much easier (relative to the leaders) games than they do when it's quarantined.

So leaving the intermingling in actually makes the disparity you mention worse.


Instead you launch into a 'if my aunty had balls she would be my uncle', discussion.

I'm not sure what you mean by this. If you have a problem with any of my arguments or conclusions, perhaps you could outline which one(s), and why?


And yes, the intermingling in round 7 did produce largish rating differentials for the top B Division players. This is what they strive for. THis is what their coaches want.

I'm talking about the fairness of prize determination. Challenge is a separate issue. And, believe it or not, there are some players at the top of B-division who want to win the prizes, as well as to play really tough opponents. There are also some without coaches.

Your own example showed that player 52 played 4 As during of the first 6 rounds. Even with a quarantined round 7, he would have played over half the tournament against much higher rated opposition. Quarantining certainly doesn't stop them getting the opportunity for games with "largish ratings differentials".

If your goal in tournament design was simply to give tough games to those who really wanted them, you'd allow players to take "ratings acceleration" (of, say, two or three hundred points) and play in division A.

ursogr8
19-04-2005, 10:49 AM
My answer was that quarantining the final round doesn't make the field less fair to those at the top - it actually makes it more fair.

As my stats should have made clear, keeping the intermingling not only allows leapfrogging, but the players just behind the leaders in B-division get much easier (relative to the leaders) games than they do when it's quarantined.

So leaving the intermingling in actually makes the disparity you mention worse.



<snip>.

The question I asked was basically this
> The SWISS pairings system has upfloats and downfloat rules, that over 7 rounds is supposed to give every one a relatively fair share of the ups and downs.
But, in what you propose, essentially players 48 and 49 get another B division opponent in round 7, when they have already had 5 B's in contrast to 50 and 52 having 2 Bs. It would seem like time for SP to give 48 and 49 the hard side of the pairings, and 50, 52 an easier side; however your SP interference (no loaded terminology intended) over-rides the in-built balancing of Dutch pairing rules. Players 50 and 52 would probably feel aggrieved that 48 and 49 get another relatively easy opponent.

starter

Recherché
20-04-2005, 04:25 PM
The question I asked was basically this
> The SWISS pairings system has upfloats and downfloat rules, that over 7 rounds is supposed to give every one a relatively fair share of the ups and downs.
But, in what you propose, essentially players 48 and 49 get another B division opponent in round 7, when they have already had 5 B's in contrast to 50 and 52 having 2 Bs. It would seem like time for SP to give 48 and 49 the hard side of the pairings, and 50, 52 an easier side; however your SP interference (no loaded terminology intended) over-rides the in-built balancing of Dutch pairing rules.

This is more or less correct. However the imbalancing effect you describe is created by the intermingling system, not my quarantine idea. And it's not a weakness - it's one of the strengths of the tournament system; players who perform well in lower divisions earn the right to increasinglly tough opposition, instead of large-differential pairings being scattergunned during the "junk" rounds.


Players 50 and 52 would probably feel aggrieved that 48 and 49 get another relatively easy opponent.

But I just explained how the pairings when left unchanged are easier relative to the pairings the top players are getting than the quarantined pairings, which are relatively harder for 49.

With relatively harder pairings for those players behind them who have faced weaker fields, and the elimination of the multi-player leapfrog, 52 and the others at the top of the field have a much fairer deal.

They should feel aggrieved at the lack of quarantining, rather than the existence of it.

Are you aware of anyone having complained about the draw when quarantining was the routine practise at Box Hill? Of course, even if they didn't, it doesn't really help my case, because likely nobody has complained about the lack of it either. Except me, of course. Now; last year when it took over; and also once when I was leading B-division and the preliminary pairings for the final round of a tournament were not quarantined - however the final pairings on the night were.

But still, it's something to think about if your main problem with the idea is that you think B-divs will complain.

It has occured to me that the paradigm-shift at Box Hill re quarantining may have coincided with Phil taking a step back. If he comes back to a more active role we should ask him what he thinks about it, and why he chose it as the system-of-choice when he was doing the pairings.

ursogr8
21-04-2005, 11:32 AM
<snip>
We'll also look at 49, the only player who could pass them to outright first.
<snip

Rob

49 was a particularly odd example for you to choose.
He was a DNP in round 3; and entered round 7 on effectively 4.5/5(played), in contrast to 52 and 57 who were 5/6.
We might try tonight to generate quarantined pairings for these players to see who is quarantined 7th round opponent might have been.
I repeat the point that 52 and 57 may have been happier with 49 getting an A opponent than a B.
But, I think we are basically looking at the wrong example, given 49's 3rd round DNP.

starter

Recherché
21-04-2005, 12:03 PM
I repeat the point that 52 and 57 may have been happier with 49 getting an A opponent than a B.

Why?

They also are getting B opponents instead of A, and the change is in their favour. In ratings terms, at least, they are gaining far more than 49 from the change.

Also you are ignoring the most important point - about the leapfrogging which quarantining prevents.


But, I think we are basically looking at the wrong example, given 49's 3rd round DNP.

I hadn't noticed that, but it's not relevant. The fact remains that 49, after having played mostly B-division opponents, was in a position to take first place and leapfrog three players who had faced much tougher fields than he had, and had performed better over the course of the tournament.

ursogr8
21-04-2005, 02:28 PM
Why?
Because they may have better fancied their chances with the scenario where they play As instead of Bs. They may have concluded that 49 would not beat an A.
In actual practice, this in fact happened, contrary to your hypothetical.


They also are getting B opponents instead of A, and the change is in their favour.
See above comment again.



Also you are ignoring the most important point - about the leapfrogging which quarantining prevents.

It is only important for your line of argument. Those on 5.5 may have reasoned quite differently...again see my above.

regards
starter

Recherché
21-04-2005, 03:58 PM
Is "the players at the top of B are probably overconfident enough to think they can win against an unfair tournament structure" the only counter-argument you have?


Because they may have better fancied their chances with the scenario where they play As instead of Bs.

That's a bit of a stretch. They'd have to be extremely confident. And to support that, all three would have to have that level of confidence as well. In fact, you'd have to be certain that in every tournament you didn't quarantine, all the players in this position felt this way, in order to even attempt to justify an unquarantined final round on these grounds. And you'd have to ask them. Otherwise you are forcing an unfair tournament structure on them without their consent. "They might fancy their chances" just doesn't cut it.


They may have concluded that 49 would not beat an A.

On what grounds would they conclude that? 49 is much closer to the rating of his A than they are to theirs. An upset with a ratings differential of only 100 points is far more likely than one where the differential is 300 or 400 points. And it's not just 3 or 4 times more likely either, because the scale is not a linear one. It's more than that.

Also, even if 49 beats his B, he cannot take outright first. This is a crucial feature of quarantining. 49 does not deserve outright first place based on his tournament performance and the tournament structure should not allow for the possibility of him getting it.

Note that even regardless of the finishing place of 49, I still think 52, 57 and 47 should be playing each other in the final round (as per the quarantined pairings I suggested) to determine final placings. The final placings have more credibility when at least some of the prize winners have actually played each other.


In actual practice, this in fact happened, contrary to your hypothetical.

The fact that it didn't actually eventuate this time doesn't mean it wasn't a real - and problematic - possibility.

Do you acknowledge that if they had all lost, and 49 had won and taken outright first, that it would have been an unfair result? If not, on what grounds do you justify it as fair, given the large disparity between fields faced, and the fact that the players above 49 never get the chance to play 49?


It is only important for your line of argument. Those on 5.5 may have reasoned quite differently...again see my above.

There is nobody on 5.5. Only 5 (47, 52, 57) and 4.5 (49), and then so on down through 4, etc.

Anyway, aside from that, if there were some players who believed that they could still do well in a tournament if they were playing black every game, does that mean a tournament structure where some players had nothing but black, and some players had nothing but white, would be fair? Of course it doesn't.

Recherché
21-04-2005, 04:25 PM
Here is a quarantined final round from the 2004 Club Championship:

http://www.boxhillchess.org.au/e2004/e0404fcc/round9.htm

On what grounds can you justify removing that quarantine, given that you know for a fact that one of the players leading B-division (me) would consider non-quarantined pairings to be grossly unfair?

If that round was not quarantined, the players below me and player #83 on 5/8 would still get the same pairings (playing each other), since there are no A-division players on 3/8. The odd one out (player 84, probably) would get an A-division player on 2.5 points, I think. In this situation it is practically guaranteed that one of them will win to reach 6/9. In fact, as you can see from the results, no less than three of those players finished on 6/9.

Now, in the meantime, instead of me and player #83, the two B-division leaders, playing each other to decide who takes first place - which anyone with a smidgin of common sense can see would the logical pairing to have - we get paired against A-division players on 3.5, of whom the lowest rated is 386 points above me and 662 points above player #83.

If we both lost, which is certainly what the ratings suggest would happen, and the other results proceeded as per the actual tournament (which is not unreasonable to assume since the pairings would be the same), then player #83 and I would be completely out of the prizes. Even if we manage to get draws we're consigned to a mass tie for first place.

Recherché
21-04-2005, 04:38 PM
Ignoring for the moment the problem of leapfrogging...

When two players are tied for the lead in the final round of a tournament and have not yet played each other, is the winner of the tournament best determined by...

a) the two leaders playing each other?

or

b) the two leaders each playing one of two players with roughly the same performance in that tournament?

Common sense and the Swiss pairing rules say it's (a), and (a) is what quarantining gets you. Not quaranting gets you (b), guaranteed, unless by some fluke there's no A-division players two points higher than the B-division leaders.

ursogr8
21-04-2005, 05:49 PM
All good posts of yours Rob.
Will look at over the w/e.
starter

ursogr8
04-05-2005, 05:44 PM
<snip>
It has occured to me that the paradigm-shift at Box Hill re quarantining may have coincided with Phil taking a step back. If he comes back to a more active role we should ask him what he thinks about it, and why he chose it as the system-of-choice when he was doing the pairings.

Just picking up the thread on this again Rob.

Yes, it clearly changed because of a personnel change. But accidental, not deliberate.

New question for you...>>> If there is an U1750 prize in the current BH Championship, will you be arguing for quarantining of U1750's in the final round?

starter

ursogr8
07-05-2005, 03:49 PM
I had an excitable conversation, with a respected authority, last evening on the Box Hill SWISS Pairings system (BHSPS). The BHSPS is not a SWISS draw and if measured against the SWISS pairings rules then clearly the BHSPS is clearly wrong.

Nevertheless, we run other tournament structures where the pairings are not made on the basis of game_score-to-date in a tournament. For example, a round-robin. And another example is an Interclub teams tourney. We don't say the pairings in these examples are wrong; just that they are different way of arranging opponents for the evening.

Perhaps the excitable argument disappears if we re-name BHSPS as the Box Hill Special Pairings System.

What do you say my friend?


starter

Thunderspirit
07-05-2005, 04:14 PM
I was the one who debated the parings last night with Trevor and Marcus Raine (sp??).

I have some major concerns. The first is that accelerated pairings are designed for very large events with not enough rounds to achieve a clear winner. This event does not seem to have that problem as so doesn't need to be accelerated at all.

The other major concerns is that players on different score groups should not play. IT IS WRONG FOR A PLAYER ON 2 POINTS TO PLAY A PLAYER ON ZERO. This is what happened last night, and insulting for the player on 2 points who should have been as high up as board 6, and the player on zero does not deserve to play someone on 2 points.

It also keeps the higher rated players playing each other when they shouldn't. No-one really likes the yo-yo of swisses but this doesn't fix it, in many ways it makes it worse.

Libby
07-05-2005, 07:18 PM
We actually employed this pairing method (or our version of it) at a recent session of the Development Squad. Our objective being to gain greatest value from the games by ensuring the maximum number of closely matched games rather than the yo-yo of some Swiss pairings.

No-one was playing for prizes etc - we just wanted some close matches.

We felt it didn't work as well as we hoped, but mostly because our field was too small and too few kids were in the significantly stronger bracket on this particular occasion.

But I did think it had merit as a way to structure an event. I also don't see that arguing over a system is best practice in the midst of an event. Debate is OK but it should be about how you might prefer to run something next time. Once a competition is underway, and assuming no-one has been "hiding" the conditions under which it will run from those participating - that's how it should run.

I've seen my daughter get completely "ripped off" in events by her final round pairing or final ranking but likewise see her finish well above where she has truly played so I think there are flaws in anything. Post event analysis is a good time to see how a system has or hasn't worked to provide the best outcomes for players. I can't imagine anyone wants to persist with a system if they don't have confidence that it gives you an outcome which best reflects the true performance of the top placegetters.

ursogr8
07-05-2005, 08:39 PM
We actually employed this pairing method (or our version of it) at a recent session of the Development Squad. Our objective being to gain greatest value from the games by ensuring the maximum number of closely matched games rather than the yo-yo of some Swiss pairings.

Our objective has to been to choose a tournament format that seems to attract a crowd of players who like competitive games as value for their entry fee. Your objective is a little different in that you are trying to get the maximum training experience.

<snip>


I also don't see that arguing over a system is best practice in the midst of an event. Debate is OK but it should be about how you might prefer to run something next time. Once a competition is underway, and assuming no-one has been "hiding" the conditions under which it will run from those participating - that's how it should run.

Libby, for my comments on the Doeberl 'U1600 v play-up' issue, I did wait until after the event. But in Lee's case, I thought his comments come from such a respected source that we could debate the core issues without it becoming too personal.


I've seen my daughter get completely "ripped off" in events by her final round pairing or final ranking but likewise see her finish well above where she has truly played so I think there are flaws in anything.

Many moons back I explored with KB my contention that B Division prizes are much more reliably 'won' by the BHSPS.


Post event analysis is a good time to see how a system has or hasn't worked to provide the best outcomes for players. I can't imagine anyone wants to persist with a system if they don't have confidence that it gives you an outcome which best reflects the true performance of the top placegetters.

And we do this anyway.....see my discussions with Recherche on this thread and elsewhere.


starter

ursogr8
07-05-2005, 08:46 PM
I was the one who debated the parings last night with Trevor and Marcus Raine (sp??).

I have some major concerns. The first is that accelerated pairings are designed for very large events with not enough rounds to achieve a clear winner. This event does not seem to have that problem as so doesn't need to be accelerated at all.

The other major concerns is that players on different score groups should not play. IT IS WRONG FOR A PLAYER ON 2 POINTS TO PLAY A PLAYER ON ZERO. This is what happened last night, and insulting for the player on 2 points who should have been as high up as board 6, and the player on zero does not deserve to play someone on 2 points.

It also keeps the higher rated players playing each other when they shouldn't. No-one really likes the yo-yo of swisses but this doesn't fix it, in many ways it makes it worse.


Lee

These are esentially the same points as you made last night.

This is not an event advertised that the pairings will strictly follow SWISS rules, and the fact that it doesn't seems to be your only criticism.

As I pointed out in my earlier posts....there are plenty of tourney formats that don't pair players on equal scores. I thought you might have commented on this counter to your concerns.


starter

Thunderspirit
07-05-2005, 11:37 PM
Firstly I hope at no time last night that you thought the debate was personal. If that happen I aplogize. Secondly, thanks for saying that I am a respected source. Many would disagree... ;)

On your point that the event isn't adviced as a swiss, and you give your pairings a funny name still doesn't really justify them. While they have reasoning behind them, it's hard to except them as a DOP when you know they are wrong.

I also know that whinging about it will do no good. As the event is not FIDE rated, a club can probably get away with what it wants... We'll have to agree to disagree...

Bill Gletsos
08-05-2005, 12:05 AM
This is not an event advertised that the pairings will strictly follow SWISS rules, and the fact that it doesn't seems to be your only criticism.However are not many of your events called the xxxxx Swiss. e.g. 2005 New Season Swiss, 2005 Winter Swiss
Now I note that they may be advertsised as super-accelerated swiss events but that term does not appear to be explained anywhere on the Box Hill web site. Permanent acceleration is not mentioned.

As I pointed out in my earlier posts....there are plenty of tourney formats that don't pair players on equal scores.Yes, but those are not run as implied swiss events.

ursogr8
08-05-2005, 07:50 AM
However are not many of your events called the xxxxx Swiss. e.g. 2005 New Season Swiss, 2005 Winter Swiss

hello Bill
Before I made my latest post to Lee I checked the printed flyer for the event (the 2005 Box Hill Championship) to see the exact wording, and to check if the entrant could mistakenly think he was entering a bog-standard SWISS. I can assure you Bill that there is clear indication this is a variant. I didn't check the Tuesday events that you quoted because Lee has not entered for these. They are much smaller fields (less than 50) and the reason for divisionalization is different from the Championship.


Now I note that they may be advertsised as super-accelerated swiss events but that term does not appear to be explained anywhere on the Box Hill web site.
The term is explained regularly by articles appearing on our notice-board; I write them. (You will recall earlier posts where I emphasized that it is necessary to get a consenus from the B Division players that they cannot get A Division prizes).
These notice-board articles are posted under the heading Bulletins on our web-site; recently there has been a re-location to the heading Archives
The term is explained in our 50 year publication.

I go back to my original comment made to Lee "This is not an event advertised that the pairings will strictly follow SWISS rules, and the fact that it doesn't seems to be your only criticism."


Permanent acceleration is not mentioned.
You are very aware that the naming of the BHSPS has been the topic of many posts over the past two years.
I have settled on some variations of Divisionalised with intermingling.
Permanent acceleration is but one of our methods to achieve Divisionalised with intermingling; but using that descriptor would run the risk of inviting B's to claim A prizes.




Yes, but those are not run as implied swiss events.
As I said at the start Bill, the flyer makes it clear it is a variant, not bog-standard SWISS. So, the implication you raise is not appropriate. And thus the point I made is a question for Lee to address. "As I pointed out in my earlier posts....there are plenty of tourney formats that don't pair players on equal scores."


starter

ps
Since you (Bill) have concentrated your sentences on only the descriptor of the event, and not the concept itself, then some readers might well presume you think the BHSPS is a good idea.

Bill Gletsos
08-05-2005, 05:22 PM
hello Bill
Before I made my latest post to Lee I checked the printed flyer for the event (the 2005 Box Hill Championship) to see the exact wording, and to check if the entrant could mistakenly think he was entering a bog-standard SWISS. I can assure you Bill that there is clear indication this is a variant.Irrelevant as you are not responding to my comments. I didnt refer to the Championship, I specifically referred to your other events where the title includes the word Swiss. From the titles a reasonable individual could expect them to be standard swiss events.

I didn't check the Tuesday events that you quoted because Lee has not entered for these. They are much smaller fields (less than 50) and the reason for divisionalization is different from the Championship.Again irrelevant to my point as I wasnt referring to the Championship.

The term is explained regularly by articles appearing on our notice-board; I write them. (You will recall earlier posts where I emphasized that it is necessary to get a consenus from the B Division players that they cannot get A Division prizes).
These notice-board articles are posted under the heading Bulletins on our web-site; recently there has been a re-location to the heading ArchivesI dont see where the term super-accelerated is decscribed in the bulletins on the website. Can you provide an actual reference.

The term is explained in our 50 year publication.Is that on the website. If so where exactly.

I go back to my original comment made to Lee "This is not an event advertised that the pairings will strictly follow SWISS rules, and the fact that it doesn't seems to be your only criticism."That comment may work for Lee, it doesnt for me, as I've criticised it previously on your competitive index thread.

You are very aware that the naming of the BHSPS has been the topic of many posts over the past two years.
I have settled on some variations of Divisionalised with intermingling.
Permanent acceleration is but one of our methods to achieve Divisionalised with intermingling; but using that descriptor would run the risk of inviting B's to claim A prizes.Perhaps but B]Permanent acceleration[/B] is indeed a feature and Divisionalised with intermingling does not in any way imply that is the case.

As I said at the start Bill, the flyer makes it clear it is a variant, not bog-standard SWISS. So, the implication you raise is not appropriate.My implication is entirely appropriate. The title of the events calls them swiss events.

And thus the point I made is a question for Lee to address. "As I pointed out in my earlier posts....there are plenty of tourney formats that don't pair players on equal scores."For other events like round robins etc it is clear they dont pair players on equal scores.


Since you (Bill) have concentrated your sentences on only the descriptor of the event, and not the concept itself, then some readers might well presume you think the BHSPS is a good idea.In which case they would be mistaken.

ursogr8
10-05-2005, 09:05 AM
<snip>
Is that on the website. If so where exactly.
.
http://chesschat.org/showthread.php?t=1268&page=1&highlight=Archives

post #10 is taken direct from the 50 year publication

ursogr8
10-05-2005, 09:18 AM
<snip> From the titles a reasonable individual could expect them to be standard swiss events.
<snip>.

Reasonable Melbourne people don't just rely on the title of an event; they read the flyer to see the T's and C's.
And we post the flyer on the local notice-board as the Arbiter's method of displaying tournament rules.
So, reasonable Melbourne entrants would not make the incorrect expectation you surmise. Of course, late entrants may make unexpected assumptions in their haste to get started. ;)

starter

ursogr8
10-05-2005, 09:21 AM
<snip>
That comment may work for Lee, it doesnt for me, as I've criticised it previously on your competitive index thread.

<snip>


Aha.
If you have made the comment on the one thread, you don't have to search 5900 posts as you claimed yesterday. Over to you to post the location that describes your position.

starter

Thunderspirit
10-05-2005, 02:05 PM
As DOP I don't really think you can justify the event by saying this is not a proper swiss even if it is adverstised before. Ok, true you can make a decision whether to play or not, and if it doesn't suit then don't play.

If the event was FIDE rated I could jump up and down, but I can't because it's an ACF event. Another club example exists in Canberra. At my beloved Tuggeranong Chess Club, players have always been able to take a half point bye in the last round. A RULE I HATE!! I hate it because its wrong under the rules. I have tried to raise it several times but I just been ignored, and got the usual but no-one on a perfect score has tried to apply for a half point bye in the last round.

Firstly this isn't true. A player on 4/4 did approach me for a 1/2 pt bye once at a Tuggeranong event, a half pt that would have won him the event. To complicate matters this was best friend Allen Mengelkamp. Allen was due to play Ian Rout a player 200 pt+ higher than Allen. I had no choice but to refuse. Not only was it dodgy, but the top players (which included Shuan) would have rioted, and not accepted the pairings. As a result Allen played Ian, and Ian won (and won the event...)

It's a club rule, but its still wrong...

ursogr8
10-05-2005, 02:46 PM
As DOP I don't really think you can justify the event by saying this is not a proper swiss even if it is adverstised before. Ok, true you can make a decision whether to play or not, and if it doesn't suit then don't play.

This sounds just a re-state of your preference. It contains no new argument.


If the event was FIDE rated I could jump up and down, but I can't because it's an ACF event.
We would always want to be under the right legalise. We are very law-abiding.



Another club example exists in Canberra. At my beloved Tuggeranong Chess Club, players have always been able to take a half point bye in the last round. A RULE I HATE!! I hate it because its wrong under the rules. I have tried to raise it several times but I just been ignored, and got the usual but no-one on a perfect score has tried to apply for a half point bye in the last round.

Firstly this isn't true. A player on 4/4 did approach me for a 1/2 pt bye once at a Tuggeranong event, a half pt that would have won him the event. To complicate matters this was best friend Allen Mengelkamp. Allen was due to play Ian Rout a player 200 pt+ higher than Allen. I had no choice but to refuse. Not only was it dodgy, but the top players (which included Shuan) would have rioted, and not accepted the pairings. As a result Allen played Ian, and Ian won (and won the event...)

^
This probably interests the casual readers because there have been few posts on CC today.
But, we do not give 1/2 point byes in the last round. Don't ask.


It's a club rule, but its still wrong...

As you have produced nothing new then we may have to move to detente.


starter

ursogr8
11-05-2005, 08:03 PM
Just picking up the thread on this again Rob.

Yes, it clearly changed because of a personnel change. But accidental, not deliberate.

New question for you...>>> If there is an U1750 prize in the current BH Championship, will you be arguing for quarantining of U1750's in the final round?

starter


Rob

Tear yourself away from the CV Newsletter a minute mate and have go at this question I posed. ^^^^

Btw, there are 11 upfloats from B Division into A division in round 4; I might start to track that metric.

Preliminary round 4 pairings are available.

starter

Recherché
15-05-2005, 07:17 PM
Just picking up the thread on this again Rob.

Yes, it clearly changed because of a personnel change. But accidental, not deliberate.

New question for you...>>> If there is an U1750 prize in the current BH Championship, will you be arguing for quarantining of U1750's in the final round?

No. Ratings prizes are not divisional prizes. They have (IMO) a different goal, and a different criteria for fairness. I would be concerned about a ratings prize that straddled divisions (eg 1200-1600), but that has never happened, and I'd be very surprised if anyone even considered it.

Are you going to address any of the points I've made earlier?

ursogr8
15-05-2005, 07:28 PM
No. Ratings prizes are not divisional prizes.
At BH, they are often the same amount of money. So how do they differ?




They have (IMO) a different goal, and a different criteria for fairness.


I thin I need more detail


I would be concerned about a ratings prize that straddled divisions (eg 1200-1600), but that has never happened, and I'd be very surprised if anyone even considered it.
Just a strawman. Not response required i think.


Are you going to address any of the points I've made earlier?

Yes.
But I need you to address the U1750 question with more explanation than just noting a naming difference.

starter

Recherché
15-05-2005, 07:33 PM
I have some major concerns. The first is that accelerated pairings are designed for very large events with not enough rounds to achieve a clear winner. This event does not seem to have that problem as so doesn't need to be accelerated at all.

These are not simply accelerated pairings.

Your assumption about the "need" for acceleration is false. The Box Hill pairing system, as should have been made clear by the rest of this thread, is designed primarily to ensure competitive games and to reward players who perform strongly in the lower division(s) with the opportunity to play opponents from the division(s) above.


The other major concerns is that players on different score groups should not play. IT IS WRONG FOR A PLAYER ON 2 POINTS TO PLAY A PLAYER ON ZERO.

Completely false. The players are in separate divisions. 0 points in A-division is not the same as 0 points in B-division.


While they have reasoning behind them, it's hard to except them as a DOP when you know they are wrong.

They're only wrong according to Swiss pairing rules. Which aren't the rules being followed. As Trevor said, a Round Robin would be "wrong" by Swiss rules as well, it doesn't mean it's wrong on its own terms.

You're free to say you don't like the Box Hill system (though I would suggest that you simply haven't fully understood it yet), but you can't wave Swiss pairing rules as evidence of it being "wrong".


This is what happened last night, and insulting for the player on 2 points who should have been as high up as board 6, and the player on zero does not deserve to play someone on 2 points.
What makes it insulting? When I was playing in B-division and started playing players rated 1600 instead of 1200 after early wins I felt rewarded, not insulted.


It also keeps the higher rated players playing each other when they shouldn't.
Why shouldn't they? Lets say I'm 1750 or so and get paired against Froehlich in the first round. I lose, which isn't surprising. Then next round suddenly I should be playing someone rated 800 who lost to a 1400 instead of someone rated about 1650 who also lost in the first round against one of the top seeds? That's silly.


No-one really likes the yo-yo of swisses but this doesn't fix it, in many ways it makes it worse.
In what way does it make it worse? If the Box Hill system is successful at anything it's successful at reducing the average ratings differential between players who play each other (and large differentials are what define the "yo-yo"). Trevor has months of data and analysis in his competitiveness thread which proves it. If you didn't realise, it's what the "competitive index" is based upon.

Recherché
15-05-2005, 07:34 PM
I think there are flaws in anything.
I think you can reasonably argue that a round-robin is entirely fair to the participants. The only flaw is the time it takes, but that's a different class of flaw. :)

Recherché
15-05-2005, 07:38 PM
However are not many of your events called the xxxxx Swiss. e.g. 2005 New Season Swiss, 2005 Winter Swiss
Yes, but Lee is specifically complaining about an event where this isn't the case. Also he's not (at least, not here) claiming that he wasn't aware of the system being used.

ursogr8
15-05-2005, 07:57 PM
Firstly I hope at no time last night that you thought the debate was personal. If that happen I aplogize. Secondly, thanks for saying that I am a respected source. Many would disagree... ;)

On your point that the event isn't adviced as a swiss, and you give your pairings a funny name still doesn't really justify them. While they have reasoning behind them, it's hard to except them as a DOP when you know they are wrong.

I also know that whinging about it will do no good. As the event is not FIDE rated, a club can probably get away with what it wants... We'll have to agree to disagree...

hi Lee

Apology about regressing to this oldish post of yours, but we have now had a good look at your claim that the event could not be FIDE rated. We now think it complies with FIDE regulations, and could be FIDE rated.


starter

Recherché
15-05-2005, 08:21 PM
At BH, they are often the same amount of money. So how do they differ?

I think I need more detail

OK, I think at this point we need to clarify what we mean by "ratings prize".

There is:

a) A ratings based prize in a monster swiss, which is partly a subsitute for having divisions (and a poor substitute IMO).

and

b) A ratings based prize in a divisionalised, intermingling tournament, which is there to give the people at the bottom (and occasionally middle) of a division something to strive for.

I assumed you were talking about (b). In the case of (a), I think divisions are preferable, and I haven't given much thought to their fairness, or how to make them more so. My basic assumption is "divisions are fairer, and more fun to boot".

In the case of (b), the ratings prizes are significantly smaller than the divisional prizes (which makes sense, since there are fewer people competing for them). See last year's Open (http://www.boxhillchess.org.au/e2004/e0409fop/results.htm).

However, upon further reflection, even in the case of (b) I'm still in favour of divisions instead of ratings prizes.

For instance, I think in the case of the 2004 BH Open, there probably should have been at least three divisions, given the size of the field.

The five divisions of the 2003 Xmas Swiss (http://www.boxhillchess.org.au/e2003/e0311fxs/results.htm) are perhaps an interesting point of discussion.

I came to the realisation while writing this post that when it comes to ratings prizes I think I'd actually argue for them to be abolished in favour of divisionalised intermingling, with the divisions small enough (say, 16-30 players) that there isn't really a "need" for ratings based prizes to break them up. But I'll have to give that some more thought before I can make a full case for it. For instance, with larger numbers of smaller divisions, it may be necessary to reduce the gap between adjacent divisions to 1 point instead of 2.

The question of quarantining in such a system is also something that would need further examination. I strongly suspect that quarantining becomes unnecessary (and undesirable) when you have more divisions with a smaller divisional separator (1 point).

It occurs to me that the need for quarantining may be a symptom of having divisions that are too large. If this is the case, then it makes more sense to address the cause of the problem (by having smaller divisions) than to treat the symptoms (by quarantining).

I shall give this further consideration. In the meantime, feel free to comment.


Just a strawman. Not response required i think.

No, no need to respond. I was just giving an example of the only sort of ratings prize I could think of that I would object to. As it turns out, I was wrong, and in fact I object to all of them! :D

ursogr8
16-05-2005, 08:11 AM
OK, I think at this point we need to clarify what we mean by "ratings prize".

There is:

a) A ratings based prize in a monster swiss, which is partly a subsitute for having divisions (and a poor substitute IMO).

and

b) A ratings based prize in a divisionalised, intermingling tournament, which is there to give the people at the bottom (and occasionally middle) of a division something to strive for.

I assumed you were talking about (b). In the case of (a), I think divisions are preferable, and I haven't given much thought to their fairness, or how to make them more so. My basic assumption is "divisions are fairer, and more fun to boot".

In the case of (b), the ratings prizes are significantly smaller than the divisional prizes (which makes sense, since there are fewer people competing for them). See last year's Open (http://www.boxhillchess.org.au/e2004/e0409fop/results.htm).

However, upon further reflection, even in the case of (b) I'm still in favour of divisions instead of ratings prizes.

For instance, I think in the case of the 2004 BH Open, there probably should have been at least three divisions, given the size of the field.

The five divisions of the 2003 Xmas Swiss (http://www.boxhillchess.org.au/e2003/e0311fxs/results.htm) are perhaps an interesting point of discussion.

I came to the realisation while writing this post that when it comes to ratings prizes I think I'd actually argue for them to be abolished in favour of divisionalised intermingling, with the divisions small enough (say, 16-30 players) that there isn't really a "need" for ratings based prizes to break them up. But I'll have to give that some more thought before I can make a full case for it. For instance, with larger numbers of smaller divisions, it may be necessary to reduce the gap between adjacent divisions to 1 point instead of 2.

The question of quarantining in such a system is also something that would need further examination. I strongly suspect that quarantining becomes unnecessary (and undesirable) when you have more divisions with a smaller divisional separator (1 point).

It occurs to me that the need for quarantining may be a symptom of having divisions that are too large. If this is the case, then it makes more sense to address the cause of the problem (by having smaller divisions) than to treat the symptoms (by quarantining).

I shall give this further consideration. In the meantime, feel free to comment.



No, no need to respond. I was just giving an example of the only sort of ratings prize I could think of that I would object to. As it turns out, I was wrong, and in fact I object to all of them! :D

Thanks Rob for this meandering, inconclusive post. It is OK to write this type of post; in fact some of my best efforts have been like that and gone nowhere. ;)

You promise to comeback and re-post after further consideration...so I think I will refrain from too much forensic analysis of your post here until then.

But, let me assist your next bout of thinking..........
our BH philosophy has been to
> have each rating prize the same (and of course, as we pursuing equity we need to make the numbers in each rating group the same). Except the top group who get a bit more.
>> use the words rating_split and divisional_split interchangeably. They are one and the same concept.

However, as your links (included in your post) point out, BH has been inconsistent in applying the first principle (>).
Clearly in the 2004 OPEN it was a better financial prospect to be at the top of B Division, rather than the bottom of A Division. I am not sure why we did that, nor that we consciously did it. Worth debating in Committee next time round.

regards
starter

Recherché
16-05-2005, 03:59 PM
Thanks Rob for this meandering, inconclusive post.

Heh. Well I was thinking over the issue as I wrote it, so it was bound to end up like that. Its present form is actually significantly tidied up from how it looked after the main writing was finished. :D

ursogr8
23-06-2005, 08:20 PM
Ignoring for the moment the problem of leapfrogging...

When two players are tied for the lead in the final round of a tournament and have not yet played each other, is the winner of the tournament best determined by...

a) the two leaders playing each other?

or

b) the two leaders each playing one of two players with roughly the same performance in that tournament?

Common sense and the Swiss pairing rules say it's (a), and (a) is what quarantining gets you. Not quaranting gets you (b), guaranteed, unless by some fluke there's no A-division players two points higher than the B-division leaders.

Rob

I don't think you can ignore leap-frogging; it is a consquence of large fields and SWISS rules. So I don't answer a) nor b).


starter

ursogr8
24-06-2005, 04:31 PM
OK, I think at this point we need to clarify what we mean by "ratings prize".

There is:

a) A ratings based prize in a monster swiss, which is partly a subsitute for having divisions (and a poor substitute IMO).

and

b) A ratings based prize in a divisionalised, intermingling tournament, which is there to give the people at the bottom (and occasionally middle) of a division something to strive for.

I assumed you were talking about (b). In the case of (a), I think divisions are preferable, and I haven't given much thought to their fairness, or how to make them more so. My basic assumption is "divisions are fairer, and more fun to boot".

In the case of (b), the ratings prizes are significantly smaller than the divisional prizes (which makes sense, since there are fewer people competing for them). See last year's Open (http://www.boxhillchess.org.au/e2004/e0409fop/results.htm).

However, upon further reflection, even in the case of (b) I'm still in favour of divisions instead of ratings prizes.

For instance, I think in the case of the 2004 BH Open, there probably should have been at least three divisions, given the size of the field.

The five divisions of the 2003 Xmas Swiss (http://www.boxhillchess.org.au/e2003/e0311fxs/results.htm) are perhaps an interesting point of discussion.

I came to the realisation while writing this post that when it comes to ratings prizes I think I'd actually argue for them to be abolished in favour of divisionalised intermingling, with the divisions small enough (say, 16-30 players) that there isn't really a "need" for ratings based prizes to break them up. But I'll have to give that some more thought before I can make a full case for it. For instance, with larger numbers of smaller divisions, it may be necessary to reduce the gap between adjacent divisions to 1 point instead of 2.

The question of quarantining in such a system is also something that would need further examination. I strongly suspect that quarantining becomes unnecessary (and undesirable) when you have more divisions with a smaller divisional separator (1 point).

It occurs to me that the need for quarantining may be a symptom of having divisions that are too large. If this is the case, then it makes more sense to address the cause of the problem (by having smaller divisions) than to treat the symptoms (by quarantining).

I shall give this further consideration. In the meantime, feel free to comment.

<snip>


Rob

I think this is one of the best of all posts in this thread.
I am nearly convinced by your argument to abolish ratings prizes altogether and replace with multiple divisional prizes. The only point where I see that this unravels in (say) a 6 division tournament is that it would be much harder to negotiate the consensus that division_2 cannot win division_1 prizes, than the current consensus where we effectivly negotiate division_4 (and _5 and _6) can't win division_1 prizes.

starter