The USCF rulebook says the following Swiss tie-breaks should be used unless posted otherwise:
Modified Median
Solkoff
Cumulative
Cumulative of Opposition
The USCF National Scholastic Tournament Regulations specify
(Individual)
Modified Median
Solkoff
Sonneborn-Berger
Cumulative
(Team)
Cumulative
Solkoff
Sonneborn-Berger
Kashdan
I realize that no tie-break system is perfect, and that at best they are a neccesary evil. However, I’m still curious about the differences. Could someone who is knowledgeable about this comment on the rationale for the National Scholastic recommendations? In particular, rule 34E8 points out the disadvantages of Sonneborn-Berger for Swiss tournaments.
Thanks for the reply, Mike. One thing that prompted my question was determining team tie-breaks (for individual-team tournaments). Adding up the individual Modified Medians does not seem appropriate, since even scores are handled differently than plus or minus scores (although this is done at many tournaments). The rule book does not offer any specific guidance for individual-team tie-breaks. It seems logical (to me) to use the recommended individual tie-breaks, without the Modified Median: Solkoff, Cumulative, Cumulative of Opposition.
The National Scholastic regulations are the only “official” recommendation I have been able to find for individual-team tie-breaks, but it is difficult to understand why Cumulative, recommended as only the fourth tie-break for individuals, should be the first team tie-break. I’m sure there is sound reasoning behind this, but I would feel better if I understood it.
I realize that for many of you this seems like a lot of hair-splitting, given that any tie-break short of a playoff is unfair. But at scholastic tournaments where time constraints do not allow for playoffs, and trophies have to be awarded, tie-breaks become a really big deal. It would help to have an official recommendation, or analysis, to point to.
I think you are looking for a level of accuracy that does not exist. Tiebreaks are inherently arbitrary. There are really only two important factors in a tiebreak system: 1) it should not be obviously unfair, and 2) it must break the tie. As to your specific question about Cumulative/Continental, the only advantage of this system is that it is fast and easy. This is a big plus if you are doing the tiebreaks by hand, and a small plus if you want the players to be able to figure out the tiebreaks before the last round.
You’re right. Unfair was the wrong word - maybe imperfect. I realize they are all inherently inaccurate, but probably some are more applicable in certain situations. It just jumped out at me that the National Scholastic tie-breaks are different from the ones suggested in the rulebook, and the team tie-breaks are very different from the individual ones. It led me to believe that someone with more insight and experience than I have must have put some thought, or even research, into it, rather than pulling them out of a hat, and I was hoping to learn something. As long as I have to use them, I would like to use the ones that are generally considered to be the most applicable, and some that I can reasonably justify. Maybe, as you said, I’m looking for something that doesn’t exist.
I think you’re giving the committee a little too much credit. For example, Sonnenborn-Berger was designed for round-robin tournaments. About the only justification I can see for using it in a Swiss is that it almost always breaks the tie, so you can get the trophies awarded before the place closes.
There is a fairly good description of the effects of various tiebreak systems on pp. 198-204 of the Rulebook. You pay your money and you take your choice.
The National Scholastics were originally organized by individual organizers, in cooperation with the USCF. I would bet that the S-B tiebreaks were preferred by one of the previous organizers, probably because he wanted to be “different” or because he had heard of them somewhere, not realizing that they were good for round-robins, but not for swisses. Thus I would bet that the S-B tiebreaks were used not for any good reason, but soley due to an organizer’s whim. Through the years, the subsequent organizers probably assumed that there must have been a good reason for S-B tb, and just kept them. Once the USCF Office took over running the events, probably no one considered making a change in a relatively minor detail like this.
It's not unusual for strange things to become institutionalized by sheer inertia.
It could be as a way to show they are independent of the USCF. Just to mimic the USCF tie-breaking rules, would not show independence. Who knows the real story, if we did would we care.
As I recall, there are few parts of the national scholastic event guidelines that haven’t been debated at length, often vigorously. I’m fairly sure that applies to the tiebreaks.
I’ve asked someone from the Scholastic Council/Committee to comment on this.
I once prefaced my announcement of prizes at an event by saying that tiebreaks are a mathematical formula designed to maximize the number of annoyed parents.
Sonnenborn-Berger may be questionable for round robins, as well.
It is intended to reward the player who has defeated the stronger (or at least the higher-scoring) opposition. But since the players are tied, and have PLAYED the same opposition, this same player must also have lost to the weaker (or lower-scoring) opposition.
For example, in a quad, if the players score:
2.0 Alpha
1.5 Beta
1.5 Charlie
1.0 Delta
– and the two 1.5s draw each other, then Sonnenborn-Berger favors the 1.5 who defeats the 2.0 and loses to the 1.0, over the 1.5 who loses to the 2.0 and defeats the 1.0.
In other words, S-B seems to reward wildly inconsistent results, and to punish “expected” results.
You are not the first to point this out. In the Oxford Companion, Hooper and Whyld write (under “Auxiliary Scoring Systems”): “Originally they were intended to supplant the normal score for all purposes. The basic principle is that wins against high scorers should be valued above wins against back-markers. (Few thought that losses ought to have a similar hierarchy.)”
Steve Immitt may be right. I’m not sure why SB is used in the order it is. I’ve asked some of my Scholastic Council predecessors to reply and haven’t heard from them yet. I’ll certainly bring up the TB issue the next time the Scholastic Council reviews the regulations.
I just received an e-mail from Sunil Weeramantry who thought that SB was used ahead of Cumulative at the national scholastic tournaments because we use accelerated pairings in some sections and that accelerated pairings hurt the Cumulative tiebreaks of those in the second quarter of the draw.
He also noted that the 11 round World Youth uses cumulative as their first tiebreak (although I think they may use a playoff for the medals - MN) and they don’t accelerate pairings. He points out that slower starters can make up the ground over the 11 rounds at the WY whereas we have events that are only 7 rounds.
When using Cumulative with accelerated pairings, you should ignore the first-round result. For example, a player with a wallchart of 1 2 2 3 in a 4-round accelerated tournament should have a cumulative of 7, rather than 8.
I can’t find this in the 5th edition, although I’m sure I’ve seen it somewhere (maybe in the 4th edition).
Cumulative should not be used in any event where computers are available to calculate tiebreaks. The only advantage of Cumulative is that it is easy to compute while looking at an old-fashioned wall chart. It has no place in any modern, serious event.
As to why the rulebook guidelines are different from the Scholastic Council guildelines - the short answer is that two separate groups wrote the two documents.
Scholastic events are subject to very different pressures and concerns (as a many-time TD in the K section of national events, I know this in spades). But, in some cases it simply boils down to “we lost this argument in the Rules Committee, but not we are the Rules Committee”.
Not quite true. Another advantage is that it depends only on the results of the tied players, whereas other tiebreak systems can also depend on the results of various third-party games, which may finish hours later than the games of the tied players. If this happens, both tied players could be forced to wait a long time before their prizes are known.
But this is probably more of a theoretical argument than a practical one.
Let’s say players A and B, each rated around 1600, each defeat opponents rated 1900 in round 1. But A’s 1900 then gets upset by an 1100 and drops out, while B’s 1900 goes on to perform well, beating an expert or two and ending up with 3.5.
With Median or Solkoff, player A is being penalized for something that’s “not his fault”, i.e. for something that happened to one of his opponents AFTER A’s own encounter with that opponent. With Cumulative, only the opponents’ results up to and including the player’s encounters with those opponents play a role, which seems more logical to me.
The main thing I have learned from this thread is that there is nothing close to a consensus. I think the basic problem is that one tie-break, say the Modified Median, might be more applicable in certain situations, but as Bill mentioned it is not really meaningful if one or more of a player’s opponents later withdraws. There is no way to know in advance which tie-break will be the “fairest” or “most accurate” in a given situation. Even if you tried to pick the best one after the fact, it could be debated endlessly.
Given that, I think the best we can hope for is uniformity, and the recommendations in the Rule Book seem to be as good a standard as any, even though they may not be the “best” in every situation.
The tie-break recommendations do not specifically mention individual-team tournaments, which many scholastics are. The simplest approach for team tie-breaks is to use the total of the recommended individual ones. But the Modified Median sometimes drops one opponent’s score and sometimes two, and I can’t see that adding them up and comparing the totals is fair or meaningful. That leaves Solkoff, Cumulative and Cumulative of Opposition as team tie-breaks, which seems as reasonable to me as any other choice. The main advantage of Cumulative of Opposition seems to be that it is a big number and is likely to break just about any tie, but that is not necessarily a bad thing. It’s a lot easier to justify than a coin flip, and may be the reason why it is included as the last tie-break.