Round robin with "too many players"

Running Chess Tournaments
21

I’m trying to learn more about this stuff, and we face a similar problem with sectioning some of our tournaments, so I’ll risk asking a foolish question of you experienced TDs.

If you do a 13-round Swiss with 18 players, don’t you end up with players either facing each other multiple times, or facing opponents of very dissimilar strength?

I’m thinking about the principle of taking 2 to the power (number of rounds) to determine how many players you can have in a section and still get a clear winner in a Swiss. 2 to the fourth power is 16, two to the fifth power is 32. So what happens in rounds six and thereafter in a 13-round Swiss with 18 players? What’s that like? Everyone gets their closest games in rounds 3, 4 and 5, then the matchups get lopsided after that?

How do players feel about that?

22

I recall reading that in either the Blue Book or Harkness Handbook a few years ago and it continued that an additional two rounds are necessary for each additional place that needs to be determined, e.g., seven rounds in a field of 32 should determine a clear first and second.

It may be another look at the tournament format is needed such as a warm up tourney held before a shorter than 13 round swiss gets started.

23

Since there are fewer rounds than players, it is always possible to pair it as a Swiss without anyone playing twice, though as been pointed our is is also possible to fall into a trap and have no legal pairings.

As a rule of thumb, two additional rounds beyond the (2 to the number of rounds) will accurately determine one additional place. AS the number of rounds increases, the Swiss approximates a round-robin.

24

Please note that you are quoting Akzidenz, not me.

25

This isn’t quite right. No number of rounds will “determine a clear winner”. What you want to avoid in a typical Swiss is 2 PERFECT SCORES.

26

Here’s an extremely simple example (even simpler than those posted so far):

colors 1 W4 L2 W6 wbw 2 W5 W1 W3 bwb 3 W6 W4 L2 wbw 4 L1 L3 W5 bwb 5 L2 L6 L4 wbw 6 L3 W5 L1 bwb
Here, there were a couple of minor upsets in round 2, otherwise everything went as “planned”, and the round 3 pairings are the only “correct” ones in terms of score. So this is an extremely feasible hypothetical example. Yet, there are no round 4 pairings.

It is a theorem that, in a 6-player Swiss, if all the colors alternate in both rounds 2 and 3, there will be no pairings for round 4.

This is an extreme example of what I call the “camp” effect. If all colors alternate in the early rounds, you have divided the players into two camps, those who started with white and those who started with black, and you have made exclusively inter-camp pairings, no intra-camp pairings. In a small tournament, the inter-camp pairings will soon become extremely scarce, and you will have to make many intra-camp pairings (those where the colors don’t work).

A less extreme example occurs with an 8-player Swiss. In this case, after 3 rounds of perfect colors, there exists only one set of round 4 pairings which continues to alternate colors, and this set is almost certainly ridiculous.

The moral? In a small tournament, make sure some of the colors don’t alternate in round 3. Things will go much better in round 4.

On how many boards should colors fail to alternate? I’d say at least 1/3 of the boards, i.e. 1/6 of the players.

What constitutes a “small” tournament? I’d say up to 20 players or so.

How does one accomplish this with a pairing program? Simply change the transposition limit for alternations from 80 to 0 (but keep the equalization limit at 200). That way, the bad colors will occur primarily in the odd-numbered rounds, where they are less harmful.

Bill Smythe

27

You may get a tournament like this uschess.org/msa/XtblMain.php … 0-11041957

In the last round the player with 8.5/9 had gone through almost the entire field. The next person in line to play him had 2.5/9.

28

But isn’t that exactly what’s supposed to happen in a round-robin? Don’t forget, the Swiss was originally invented to enable tournaments to be held when there were too many players/not enough time for “real” (round-robin) pairings.

29

That’s nowhere near the limit. I once had to give the tournament leader a BYE in the last round. (and, yes, it WAS correct).

30

Not exactly a dramatic finish to a 10-round event, eh?

But the Swiss format got you around the problem that half your field had gone to bed by round 10.

Thanks, all, for the insights. Lots to think about.

31

Sorry about that. In trying to save space I should have deleted the outer quote instead of the inner one.

32

Changing the transposition limit isn’t likely to have the desired effect. That only makes it more likely that bad colors will stand, but in such a small tournament, there aren’t going to be many opportunities to transpose pairings anyway (if you’re still trying to maintain score groups).

With things like 4 rounds for 6 or 5 rounds for 8, the hybrid RR-Swiss is likely to be the simplest and safest way to handle the section.

33

Bingo!

Nothing will trigger player complaints faster than the TD making small changes to the colors/pairings that the players don’t understand. Even when the TD is “right”. The best TD looks for a generic method, supported by the printed rules, that handles the problem withOUT cleverness on the part of the TD.

Personally, I know that under extreme stress my ability to be clever goes way down. I’m a lot smarter when I do the analysis in advance and simply avoid the obvious pitfalls.

I’m a big fan of the hybrid Swiss-guided RR. I’ve used it many times, and it has never failed to produce a result that was: a) fair, and b) easy to explain to the players.