Quads - swiss vs round robin

Checking some directors’ tournament history, have notice a difference in the reporting of the quad results. The standard quad is with four players, each player plays one game with everyone in the quad. How the director reports the results, could be in a swiss or a round robin. The standard, in my area has been as a swiss than a round robin.

If the swiss or the round robin format can be used. Would one be more positive in the post tournament rating than the other. As the quad in a swiss would be three rounds. If the quad is in a round robin, it would be four rounds. If the K factor change with the amount of rounds, would not the swiss or the round robin have different post-tournament ratings? As the K factor for a 1400 player would be at thirty-seven as a three round event, as a four round would be at thirty-six. The change in the post-tournament rating, even with a difference of one point in the K factor is minor. It is still a change in the K factor, making the post-tournament rating have different ratings.

The K-factor does not depend on the number of “rounds” – whatever that means. Rather, each player’s K-factor is determined by the number of actual games (not byes or forfeits) played by THAT PLAYER.

In a quad, each player plays 3 games, regardless of whether it is reported in Swiss format or round-robin format. So it comes out the same either way – each player’s K-factor is the one that is used for players with 3 games.

If, in a large 5-round swiss, one of the players drops out after 3 games, his K-factor is based on 3 games, while everybody else’s is based on 5 games.

Bill Smythe

Bill is correct, the K factor is based on the player’s pre-event rating (including the number of rated games the person has played previously) and the number of ratable games the person played in this event.

See pages 8 and 9 in the following:
math.bu.edu/people/mg/ratings/rating.system.pdf

It makes no difference to the ratings process whether a quad is reported in round-robin format or as a swiss. The order in which the games were played is also irrelevant.

In the USCF rating system, the choice of color also makes no difference.

FIDE has started asking for the color assignments on FIDE rating reports, though I don’t know what, if anything, they’re doing with that information. Perhaps nothing, at least for now. (I don’t think that information is REQUIRED yet, though that, too, could change.)

I think USCF should begin asking for color assignments, as well – not for ratings purposes (although that would be a future possibility, I suppose), but because that information could be added to the MSA crosstable, presenting more information of interest to those who like to browse crosstables.

Bill Smythe

The new database structure has a field for color by round, I just haven’t figured out the best way to make it part of the upload data.

Thanks Bill.

The major problem I have with the conversion of the round robin into a swiss, is the pairings after the event. In my firm faith, if the tournament was a round robin, it should be reported as a round robin not as a swiss. As the pairing numbers of the round robin, at the club level is designed with first registered as pairing number one, down the line till the last player registers for the event. There is not a problem, if the final crosstables of the round robin are set up with the higher score, if a tie in score the higher rating for the higher pairings.

The problem I have with the conversion to a swiss, it is not the problem of the higher score or tie in score the higher rating. It is the question of the pairing assignment of the rounds. If a player has an upset, would not the player have a better post rating if it was in round one or round two, or in any other round.

Going from the round robin of first registered to last registered, conversion into a swiss with higher final score would give the director an open assignment for player A vs. player B in any round. If you win a game, does it matter if the assignment of the round be in any round? Would not that change the post tournament rating?

It may not still be the case, but Swis-Sys used to convert round robin events into swiss events for reporting purposes.

Having written the current ratings computation programming, I give you my assurances that it DOES NOT CARE if the event was entered as a Swiss or as a Round Robin!

In all the years I have been in unrated blitz events. It has always been without a pairing program. If the local club does have a rated blitz event, very sure it will be a round robin event without a pairing program. Even the manual changes from the first come first pairing, to the final round robin crosstable of highest rating, human error can happen. It will show up on the proof, but the corrections would not take long.

The manual conversion from round robin to swiss, human errors can happen. True, the swiss will have everyone play against each other, but what order player A plays against player B. Give you an example, if I change the swiss rounds around. Change the results of round three to be round two, than change the results of round two to round three. Can you say the post rating would not change. Can you say the order of games does not matter, if the correct orders of your opponents’ ratings are round one 1800, round two 1650, round three 1600 and round four 1700. If it is reported as round one 1800, round two 1600, round three 1650 and round four 1700. Are you sure the order of the rounds does not matter?

If it does matter, this simple problem can be a problem of the conversion of the round robin into a swiss.

Just a reminder… When we first submitted tournaments online, ‘swiss’ was the only type supported. it’s easy to convert a 4 player quad to a swiss - but it does get more difficult as the number of players increase. All the checks that are built in to help the TD get the results right would ensure the accuracy of the report. And afterwards the ratings are going to be same regardless of the way the tournament was entered.

That is the point, if the round robin was with twenty players, the conversion into a swiss can be a problem for any director. This was the problem I wanted to talk about in blitz tournaments. Clubs that have blitz tournaments, would be rare to none to use a computer pairing program during the club meting. If the tournament was a round robin, it should be submitted as a round robin. The conversion into a swiss, even if the results are right, the final crosstables can be random. As player A could be assigned to play player B in any round, making the crosstables random. The only reason why the director would convert a round robin into a swiss, as the director does not get any credit for round robins but does get credit for a swiss.

Nope. The order of a player’s games in a tournament has no effect on the post-tournament rating. All that matters is the ratings of each opponent, and the player’s total score against those opponents.

In fact, it does not even matter WHICH players you defeat and which defeat you. If you win against an 1800 and lose to a 1300, your rating will come out the same as if you lose to the 1800 and beat the 1300.

Again, nope.

Apparently you are under the misconception that ratings are calculated one game at a time, with a player’s post-tournament rating from one game being used as his pre-tournament rating in the next. THAT’S NOT HOW IT’S DONE. Instead, the player’s pre-tournament rating is used for ALL his calculations.

Let’s say a 1500 with a K-factor of 30 first defeats an 1800, then loses to a 1500. From the table for K-factor 30 on page 267, we find:

  1. 1500 defeats 1800 and gains 25 points.
  2. 1500 loses to 1500 and loses 15 points.

His net rating gain is 10 points.

Now let’s say it happens the other way around:

  1. 1500 loses to 1500 and loses 15 points.
  2. 1500 defeats 1800 and gains 25 points.

His net rating gain is still 10 points.

Bill Smythe

Thanks Bill.

What if it is a double round robin. If the player wins the first game than lose the second game. Would not the conversion from a round robin to a swiss be of a problem. If the first game is placed into section two, than the second game is placed into section one. As double round robins the players would place (2.0, 1.5, 1.0, 0.5, 0.0) the total score. If the director sees scores of (1.5, 1.0, 0.5) the director would not know the order. If you won an upset in the first game, but reported in section two. Would not be a problem for the post-tournament ratings.

Doug,

Double play tournaments are rated easily by USCF as double round robins or as double play swisses (not two sections). The conversian by the TD is easy.

Tim

Doug, I think you’re more concerned with experience requirements for advancing to the next TD level, something you seem to be fixated on.

That has NOTHING to do with the way the ratings system handles an event.

Performed a test tournament, the use of the (# = 1.5, $ = 2.0) did not work with the swiss. It should have worked with the round robin, but it added the results up, but was reported as an error in the validations.

If the conversion of a double round robin into a swiss, it would limit the section to eleven players. As a double round robin with eleven players would be twenty games, the amount of games in a section is limited to twenty rounds. The director could have the same player in the same section more than once. This would be non-standard, but it would let the player have more than twenty-one games in one section. The K factor table on page 265, only go to twelve rounds. What its’ the limit of rounds to the K factor?

The reason I ask, if a player plays a little less than one-hundred games, the player could be reported in the same section five times. The players do not have to be placed into five different sections of twenty rounds. The player could be placed into a section five times in a section of twenty rounds.

If the K factor has a set limit of rounds, would it not be best to have more than one section?

Already have the experience requirements to take the senior directors test.

Okay, okay, I’ll have to concede you have a point. If a double round-robin has so many rounds that it must be split into two events for reporting and rating purposes, then it could make a difference which games go into which half of the rating report.

But, even then, you have it backwards. If a player is trying to increase his rating, he would prefer to have his poorer results rated first, and his better results last.

Let’s say a 1500 with K-factor 30 plays in a 5-round event and gets draws against five 1800 players. The table gives him 10 points per game, for a rating gain of 50 points. His rating is now 1550.

Then he plays in another 5-round event and gets draws against five 1200 players. Since the rating difference is now 350 instead of 300, he loses 11 points per game instead of 10, for a rating loss of 55 points. His rating is now 1495.

If it had been the other way around, he would have first lost 50 points, then gained 55, for a new rating of 1505.

Bill Smythe

Also, Doug, I get the sneaking suspicion the reason you are so interested in the concept of dividing a tournament into two for rating purposes, is to get a leg up on your next level of TD certification.

If you get credit for two tournaments when you really directed only one, you’ll get there twice as fast, right?

Bill Smythe

I’ve got most of the hooks in place for double-round-robin events, at least those entered online. Based on Bill’s suggestion from a few months ago, I’m using the first two characters of the 5 character result/opponent field, so you’d enter results like:

WW
WL
DD
WF (2nd game is a forfeit loss)
etc.

I need to make sure I’ve got all the checks in place and create some test events before I’ll activate that feature, hopefully later this year.

Also, the internal version of the data entry program that they use at the USCF office can now handle up to 32 rounds, I’ll port that code over to the TD/A version one of these days. As far as I know, neither Swis-Sys nor WinTD will create events with more than about 14 rounds in them, and definitely not with more than 20 rounds, so for now events that long are going to have to be entered online as it’s not possible to create an upload file for them.

Do I need to worry seriously about events longer than 32 rounds (including matches) or round robins with more than 32 players in them?

The old USCF program Tnmtadmt has a limit of 20 rounds with the instruction: “If section has more than 20 rounds create another section.”