I wonder if anyone (but preferably an experienced TD) can help me to understand the intricacies of tiebreak systems used in USCF Swiss tournaments.
I do not have USCF rule book handily available, google search pointed me to wikipedia article (how reliable is it?), which indicates that preferred USCF tiebreak order is :
1.Modified Median
2.Solkoff
3.Cumulative
4.Cumulative opponent’s score.
The article goes into some detail to describe those methods.
I am guessing that preferred order doesn’t mean TD has absolutely follow it.
I am also guessing that TD is dependent in many cases of pairing software that is used to run the tournament.
Now, my question did not arise just from curiosity, my son just played in the State event where he scored
3.5 /5. Another child, who my son drew in the last round also scored 3.5 /5. Both kids shared 5-6 places, but tiebreak put one of them on the fifth and the other on sixth and I am trying to understand the rational.
I have results sheet in front me and it looks like this:
Points TBrk1 TBrk2 TBrk3 TBrk4
5. Player 1 3.5 11.5 13.5 11.5 42.5
6. Player 2 3.5 13.0 15.0 10.5 46.5
It is worth to mention that the pairing software is Windows TD version 4.11c 2008.
I would greatly appreciate if someone takes time to de-cypher those numbers for me.
True. But Rule 34E states that “Unless a different method has been posted or announced before the start of the first round, players will expect the following sequence of tie-break systems to be employed as the first four tie-breakers.” (It then lists the four systems you did in the order in which you listed them.)
I can’t think of any reason why a TD should have to be dependent on the pairing software used, since it should be possible for him to evaluate each tie-breaker manually (though in a large tournament which had a lot of ties, this might not be practical).
Each of these systems is an attempt to look beyond the final scores and evaluate reasons why it may be more impressive that one of the players achieved that score than it is that the other achieved that same score.
The Modified Median system, for example, sums the final scores of the opponents each player faced and then discards the highest and lowest of those scores. Presumably, if the final scores of the opponents one player faced are higher than those of the opponents the other player faced, it is more impressive that the first player achieved that tying score than it is that the other player achieved that same score.
The Solkoff system works the same as the Modified Median system except that no scores are discarded.
The Cumulative system totals the scores each tying player had at the end of each round. The rationale behind this system is that one player might, e.g., achieve a 3.5 score by losing a game, drawing a game, and then winning 3 games while another might achieve it by winning 3 games, drawing a game, and then losing a game, and the second is likely to be the stronger player, since he was presumably being matched against stronger players when he drew and lost as a result of his initial wins, while the first was presumably being matched against weaker players when he scored his wins at the end due to his initial loss and draw. The first player’s scores would be 0+0.5+1.5+2.5+3.5=8.0, while the second player’s scores would be 1+2+3+3.5+3.5=13.0.
I would guess that, in this case, it was the Cumulative system that was used to determine the placings, since it appears that Player 2, who was given the lower placing, came out ahead using the other three tie-breaking systems. The Cumulative system is frequently used for large tournaments, since it is the fastest and easiest to use, and since it doesn’t require waiting until all of the opponents each player has faced at the tournament have finished their final round games.
Slight correction:
Median discards the highest and lowest scoring opponent.
Modified median discards the lowest scoring opponent for players scoring >50% and the highest scoring opponent for players scoring <50% (both for players scoring 50%). So player six had the best four opponents scores add to 13 (11.5 for player 5) and both players had a 2-point opponent added back in for Solkoff (15 and 13.5). Opponents’ forfeits and withdrawals throw in some complications, but they don’t happen that often.
P.S. Tiebreaks are designed to try to determine which player had the more difficult road to travel in reaching the tied score (generally the more difficult opponents - with MM and Solkoff basing that on how the opponents did).
P.P.S. Every tie-break system is unfair, and decent ones are chosen to be unfair in a blindly unbiased manner.
Unfortunately, it is impossible to tell what that means given the information provided. We’d need to know the settings on the program as to what tiebreaks were selected so that we could explain. It is odd that the TD didn’t set tiebreak1 to be the most important.
Recently while playing around with WinTD I noticed that no matter how I set the tiebreaks the Tiebreak 1, Tiebreak 2, … list never changed. The defaults I set worked fine when I used WinTD’s Reports → show Prize List feature. When I used the edit window to change the layout, after clicking on the P icon, (where tiebreak 1, tiebreak 2, … are listed generically) I found I could change the Tiebreak order there to where I set them earlier (File → edit info); i.e., I moved the generic Tiebreak 1 (etc.) from the Available to the Current screen in the order I wished. The trick is that you have to know what tiebreak 1 (etc.) represents on the File → edit info screen. Because I did not do another experiment, nor was I in the mood to see if I could use some setting the make Tiebreak 1 (etc.) match my choices, I shut down WinTD and went about my business elsewhere. I suspect if a TD is using one of the “define style” settings and those tiebreak 1 (etc.) is already listed they could simply change the order on the Current screen.
At any rate that is a long winded explanation of why I believe things like that happen. It simply appears that the tiebreak choices the TD chooses to hand out prizes does not match the generic Tiebreak 1 (etc.) labels. Does anyone know an easy fix for that? So, printouts with the Tiebreak 1 stuff may not always match the the print out order in the “Show Prize List” print out (which actually names the tiebreak system).
It seems entirely possible that all four tiebreaks are, in fact, the recommended ones, and are in the recommended order, but that the listing for some reason was not in tiebreak order.
For example, Modified Median (presumably TBrk1) should be expected to come out a little smaller than Solkoff (presumably TBrk2) because of the omission of the lowest-scoring opponent. Cumulative (presumably TBrk3) could come out anywhere between 8.0 and 13.0 for a player scoring 3.5 out of 5. Cumulative Opponent’s Score (presumably TBrk4), being a composite calculation, would be expected to come out much higher than the others.
Were trophies involved? If so, it’s possible the TD simply overlooked that the printout was not in tiebreak order.
Did either kid also win another trophy (e.g. 5th overall plus best 4th grader)? If so, there might have been a limit-one-trophy-per-player policy.
Being seasonal chess player myself, I have the exact same attitude to this and I wanted to keep my son out of this conversation. At this age, he should be concentrating on improving his chess skills and not on trying to win some battles outside the board.
I have not approached the TD at the tournament, but I have sent a very polite e-mail to him asking to clarify the tie-breaks used. If he replies, I will summarise his response here.
Of course, you can use ‘Performance of the Opposition’ and just use one tie-break system.
Most of the other tie-break systems are just simplified methods to guess who played the toughest schedule in the tournament.
Pre-pairing programs, it was too complicated to calculate ‘Performance of the Opposition’ directly so various simplifed hacks were developed to approximate it.
In-born human conservatism being what it is, most old-line TDs/promoters/coaches are married to these obsolete calculation systems. If tie-breaks were invented tomorrow, no one in their right mind would develope or use the current recommended tie-break order.
TD has not replied yet, but tournament results are now available online.
If this not too much to ask, can someone check the last tournament of this player (14314783) to see whether 6th place on tie-break for hime is in fact fair result.
Looking at the tournament it depends on the tie-break order. main.uschess.org/assets/msa_joom … 2-14314783
Median, Modified Median, Solkoff and Sonnenborn-Berger are all in player five’s favor while cumulative is in the favor of player six.
Dear Mikhail - From what you have described, the only unfortunate thing is that there were exactly five trophies. If there were only four trophies, your son would be just as proud as the player he drew in the last round, and they would have shared 5th and 6th. Your son won three games, as against one draw and one loss - bravo!
When I started out in tournaments, in 1972, ratings at the beginner end of USCF’s rating scale were more in the region of 1000 to 1200, and my first rating was 1101, after scoring two wins and seven losses in my first tournament, which was a junior tournament in Manhattan.
A few months later, my rating was published (this process took several months back then!). When I lost to a much more experienced player on first board for my high school in a team match that fall, my older opponent asked me if I had a rating, and what it was: I told him “1101, but it’s going up.” “What is it going to be?”, he asked. I proudly replied “1170!”
Years later, when both of us had reached the master level, he never stopped enjoying reminding me of this first meeting between us.
May your son have many years of progress and enjoyment!
Bill, I ran the back room for a fairly large scholastic (at least for North Texas, of about 150 or so kids), several
years ago, and the thought of some of the tournament directors was that to arrive at the tie-break winners
one would look to see how many “columns” were led by each of those tied. A thought that was thankfully
and quickly corrected. But, I wonder, could the same have happened, perhaps in this case??
For the benifit of everybody who replied or followed this thread, I have an update.
TD from the event replied back to me that there was an error in results in the award ceremony and Leo did in fact placed fifth on tie-breaks. TD will have to order another trophy and asked whether I will be willing to come and pick it up, to which I gladly agreed.
Bravo to the TD. If one runs enough tournaments, every now and then, you vision can get cross-eyed,
and read the results incorrectly. I think most of us at some point have goofed up at least one award.