I received the e-mail below from Michael. My answer was that there is no “official” way to do what he is asking about; however, I think one of you probably have a better way than the method he describes for creating a “ratings balanced” section.
Tim
My questions were around the pairings process for a
recent tournament.
The tournament advertised “ratings balanced”
sections for arated 10 Minute Championship qualifying
round. 21 players registered andwere divided into 3
sections by taking the top 3 highest rated
players,seperating them in sequence (1-2-3), and then
proceeding down the list ofremaining highest rated
players and grouping them the same way, 1-2-3.
By grouping the sections in this fashion it created
an imbalance inthe average rating per section. (If the
highest rated player is always atthe top of the list
to choose from when the sections are divided,
1-2-3,then section 1 will necessarily be of a higher
average rating than 2 or3) …Is this the
proper way to divide “ratingsbalanced” sections for a
USCF rated tournament??
My other question comes from how a 22nd player was
added into the mixupon arriving AFTER the 3 sections
were already finalized. The TD simplyadded that player
to the third section causing further imbalance. What
isthe appropriate manner to add in a 22nd player at
that point?
I had included each of the section’s actual ratings
averages lastweek and can only recall partially, but
they came out to about 1760 forsection 1, 1650 for
section 2, and 1560 for section 3.
Any help would be appreciated.
As to late-arriving players, I suppose you could just continue the pattern, even though the new additions aren’t the lowest-rated players. It’s as good as any, I guess:
It’s hard to invent something new. This (Bill’s version) is the old Holland System, used for the Olympiads (and the U.S. Open) before the invention of the Swiss. It’s covered in the old Harkness rulebook. Used to be on the NTD exam, but they may have dropped it. It does an okay job, though there were always complaints about the seedings.
The help file describes the algorithm for creating balanced sections:
“If the grouping is to equalize average ratings, WinTD determines how many sections are needed to accommodate all the players in sections no larger than the requested size. For instance, with 37 players and a desired size of six, it will make seven sections. Five of these will have five players and two will have six. The players are sorted by rating, and the highest rated players are assigned to the sections in order. The next highest rated players are assigned in reverse order. At that point, the average ratings are computed for all the sections. The next available player will be assigned to the section with the lowest rating. The next after that will go into the section with the second lowest rating. This will continue until all sections have three players. The average ratings will again be computed, and the fourth players will again be assigned with the highest rated player remaining going into the section with the lowest average through three players.”
Judging from that help screen, the WinTD method is a super-sophisticated version of the zig-zag method. It sounds impractical to do it that way by hand, if time is of the essence (as it always is during registration), but for a computer it’s a snap.