Tournament format
Hank Youngerman
that would best ensure that the final results were reflective of
skill. The format should assume a tournament that might take two days
or more of play to complete.
Two things are obvious:
1. Any format should be seeded, to ensure that the better players are
not concentrated in any one competitive group.
2. The more backgammon that is played, the better.
(Before I go further, let me say that I'm somewhat but not
overwhelmingly interested in considerations of convenience and
logistics, and that answers should be given without respect to match
length, but only in relative terms. That is, if you assume that a
single 15-point match takes as long to play as (say) 4 5-point
matches, that should be a consideration in choosing between two
formats that would employ either one long match or four shorter ones.
But either format could be based on longer or shorter matches, with
similar changes in effectiveness.
Let's say you are playing a small local tournament. You choose to
play single elimination with random draw and 5-point matches. It
doesn't really matter that much that the best players will not always
win, or even that a really weak player might win now and then. Over
time, the better players will win the most and that's what you sort of
hope for.
I'm not looking for something with drama, test of mettle, or the like.
Single elimination has that drama and tension throughout, and you can
never complain that something happened in a match between two other
players. But it might not provide the highest chance of the best
player winning. Even in a seeded format, there is enough luck that a
clearly better player could lose to a weaker one in a medium or even
long match. You now have the best player sitting on the sidelines
while the tournament continues, when he could be demonstrating his
skill.
I'd like to make a suggestion as to a format, but I'm posting this as
fodder for thought and to stimulate discussion. I'm going to assume
you have 64 players, and six sessions to complete play in. Clearly,
one alternative would be a single-elimination bracket. Is there a
better way?
You begin by dividing the players into groups of approximately equal
size and strength. In each group, you play a round-robin. A certain
number of players from each group qualify to the next round. A
percentage of the score is carried over, say 50%
In the second round you again play a round-robin, with a percentage
carryover, and then a third round-robin. Eventually you come down to
four players. They play a round-robin of longer matches, with the top
two qualifying for a finals, with carryover.
For example:
Start with 64 players, in 8 groups of 8. Play a round-robin of
5-point matches, 7 in all. Qualify the top 4 in each group. In case
of ties, perhaps have a tiebreak procedure if there is a tie at 3-4
but guarantee everyone with a record of 4-3 a place in the next round.
Now, take 32 players in 4 groups of 8 (if one or two groups has 9,
this isn't a major problem other than scheduling). Cut to 16 players,
with scores being matches won in the second round plus half the
matches won in the first round. Bear in mind that some of the
randomness of the strength of the field is being somewhat quickly
eliminated.
The next round might have longer matches, say, two groups of 8 playing
a complete round-robin of 9-point matches, qualifying only the top 2
from each group. Again, score is half the score after the previous
round plus score from this round. So first-round matches by this
point are counting 1/4 and second-round matches counting 1/2; also
ties are becoming far fewer.
The final four play a round-robin of 9-point matches. The top two
qualify for the finals. The finals might be a 19-point match or
thereabout, BUT the player with the best cumulative score in the
qualifying rounds starts with a score equal to his margin over second.
For example, let's say players A and B won the following:
A B
Round 1 5 6
(carryover) 2.5 3
Round 2 4 5
(carryover) 3.25 4.00
Round 3 5 5
(Carryover) 4.12 4.50
Round 4 3 1
Total score 7.12 5.50
In the fourth rond, player A won all three matches while the other
three players each won and lost one against the other two opponents.
Player A of course qualified for the finals, player B advanced since
his carryover from previous rounds happened to be highest of the three
others. In the finals, player A starts with a 1.62 point advantage
over Player B, rounded to 2, so the finals starts 2-0 to 19.
Is this complicated? Well, probably more difficult to understand than
to actually do. I hope what I've suggested is vaguely clear. The
real question is - is this more likely to produce a situation where
the best player wins, the second-best player is second, etc., than six
rounds of single-elimination play where each match takes as long to
complete as would a round-robin of 7 5-point matches?
A second question. In a single-elimination bracket, which is better:
1. A single match to X points; or
2. Best 2-out-of-3 or 3-out-of-5 of matches to Y points;
where the expected total time to complete each option is the same.
Just looking for ideas and input - thanks.
OSMAN
>
> I'm curious as to what people would suggest as the tournament format
> that would best ensure that the final results were reflective of
> skill. The format should assume a tournament that might take two days
> or more of play to complete.
>
> Two things are obvious:
>
> 1. Any format should be seeded, to ensure that the better players are
> not concentrated in any one competitive group.
>
> 2. The more backgammon that is played, the better.
>
For 64 players, I would suggest a 6 to 8 round Swiss tournament with 7
point matches at each round. This guarantees that everybody plays 6 or
8 matches regardless of performance, and in the last several rounds,
they play against players who are at a similar perfomance level with
each other.
In case you are not familiar with the Swiss system:
1- sort all 64 players according to their rating (this is why you need
an accurate rating system)
2- divide the group into two halves from the center and pair up the
first round in such a way that player ranked 1 plays against the one
ranked 33, 2 against 34, 3 against 35, etc.
3- Once the first round matches are finished, line up the winners (again
rating order), divide the group into two; pair them up like you did in
the first round... similarly , pair up the losers. There are rules to
deal with the odd playes, etc. that I will not get into details at this
point, but let me know if you need these details. Note that two players
cannot play against ech other more than once.
4- At the end of the second round, pair up all the players who won two
matches; then pair up those who won only one; then pair up those who
lost both matches.
5- Proceed in a similar fashion with pairings until all te rounds are
done.
This system was created to identify a clear winner from a large group of
participants in a few rounds. It has been tremendously successfull in
chess tournaments and has took over the traditional round-robin system
for most open tournaments.
Hope this will add to your library of tournament format ideas.
Cheers..Osman
--
Osman F. Guner
os...@prodigy.net
http://pages.prodigy.net/osman
Julian
<hankyou...@home.com> writes
>
>You begin by dividing the players into groups of approximately equal
>size and strength. In each group, you play a round-robin. A certain
>number of players from each group qualify to the next round. A
>percentage of the score is carried over, say 50%
>
>In the second round you again play a round-robin, with a percentage
>carryover, and then a third round-robin. Eventually you come down to
>four players. They play a round-robin of longer matches, with the top
>two qualifying for a finals, with carryover.
>
the problem I foresee is that round-robins can be difficult to keep
organised - if players finishing games pick their next opponents at
random from those currently idle, you can end up with a backlog of games
and players unable to find a fresh opponent. Also, slow players can
cause horrendous problems when they have to get through 7 matches. An
alternative would be within each group of 8, play it like a four-round
mini-Swiss. If the top 4 go through, then good players have to lose at
least twice, usually 3 times to get knocked out. There's enough slack in
the pairings that you don't end up waiting for two individuals to be
free to pair up for the last round, and enough play to make sure skilful
players will (probably) make it through to later rounds.
--
Julian Hayward 'Booles' on FIBS jul...@ratbag.demon.co.uk
+44-1344-640656 http://www.ratbag.demon.co.uk/
---------------------------------------------------------------------------
"Over two-thirds of 11-year-olds failed the standard reading assessments
this year. The Government has pledged to increase this to three-quarters
by the year 2001..." - Classic FM News
---------------------------------------------------------------------------