Match Play Question

Running Chess Tournaments
1

If only one player chooses an opponent is it still considered Match play?

We are planning something different next Monday Night. I had considered what we are doing
would have to be submitted as Match play and so follow those rules.

I realized that only one player will be choosing. The plan is to put all the names in a hat
or something. A name is drawn and they choose their opponent within the 400 point range etc
If at the end of choosing there are people left over I create a seperate section that is paired
like a standard swiss. (Top vs Bottom)

But rereading the US Chess Match rules e. … (ie, the two players chose to play each other). …
it makes me wonder if this concept truely falls within the Match play boundaries.

When asked to do an event like this before I’ve always turned down the idea because of the Match play
limitations.

This is what I posted on our Web page.

Next Monday August 21st - US Chess Rated Match play Game 90;d5

US Chess Match Play Rules

These are not the complete set of rules but are items to be aware of
when choosing your opponent etc.

Both players must have established and published ratings, and those ratings must be no more than 400 points apart as of the most recent published ratings.

A player may gain or lose no more than 50 rating points in a match. (Not a problem with a one game match.)

Any player who playes in a match and who is either at his or her floor or reaches his or her floor
as a result of that match will be consider to have submitted a request to have that floor lowered by 100 points. The USCF ratings department will review
the player’s tournament history to decide if the floor should be lowered.

2

Do the players play multiple games against each other once the lineup has been selected? If not, it probably isn’t a match, because that usually requires that the two individuals play two or more games against each other. There have been some attempts to find new ways around the match rule limitations, but if someone looked at the crosstable for your event, would it be obvious that some of the players were choosing their opponents?

Remember, the point of match limitations is to preserve the mathematical integrity of the ratings system by ensuring that players have a large pool of opponents. (In the extreme case, if all of player A’s games were against player B, how reliable would player A’s rating be?)

It doesn’t sound like this event poses that kind of mathematical risk to me.

3

It is just one round of game 90;d5 for the night. We have an inherently small pool naturally on Monday nights regardless.
As stated above I viewed what we were going to be doing as falling under the Match guidelines because of the player choosing their opponent. It wa only the realization that half the field would have no choice that made me wonder.

This is probably going to be a one off experiment. :slight_smile:

4

Players who play most of their games with a small pool of opponents have the potential of creating reliability issues in their ratings, but that’s not an easily solvable problem. What the match limitation rules try to do is deal with the more extreme of these types of situations.

5

Is this not a little like a ladder event? Will the USCF rate a ladder event where over the course of many multiple nights two opponents may play each other more than once?

Larry S. Cohen

6

I agree it does bear some similarity to some ladder events, and ladder events are ratable, though I doubt anyone could tell from the crosstable records how many of them are run. I have my doubts that anyone could easily pick this event, if held, out of the crosstables as one raising issues of statistical reliability, either.

Prof. Elo probably never put it this way, but as long as players follow the law of large numbers (ie, play a sufficient number of different opponents), the ratings system doesn’t care who they played or why it was paired that way. (Note that all of my comments are my personal opinion, not US Chess policy.)

7

Around 1994 to 2000 we ran a few “Choose Your Opponent” events at the club on Lunt Avenue, and US Chess rated them OK. I get the feeling, though, that this sort of thing may be scrutinized more carefully nowadays.

Our setup worked as follows:

The tournaments were four or five rounds, just like our Swisses.

In the first round, beginning with the top-rated player, each player would choose an opponent from among those not yet paired. No player had the right to refuse to be chosen. The player doing the choosing would also choose color.

After the pairings were thus determined, and the games started, each player would be given 1 “choice point” if he was the chooser, or 0 “choice points” if he was the choosee.

In subsequent rounds, the players with the fewest total choice points would choose first. All choices would be made without regard to score, color, rating, etc. In other words, you could choose any player you hadn’t already been paired with, from among those not yet paired for that round.

In any round, a player asked to make a choice would have the option to pass. Then that player could be chosen by somebody else. Passing might be a good option if the player you wanted to play has already been paired. By passing, you would get 0 choice points for that round, increasing your chances of getting your choice in subsequent rounds.

After all players have been paired by this choosing process, there may be a few players left over who passed and were then not subsequently chosen. These leftover players would be paired arbitrarily (or possibly 1-vs-2-style by rating) by the TD. In this case both players in the pairing would get 0 choice points.

If you want to try this idea, I’d suggesting having no or very small prizes, or possibly gag prizes like “bravest choices” or “biggest upset by a chooser” or “biggest upset by a choosee”.

Also be prepared to (possibly) have this event rejected for rating on the grounds that some players chose their opponents. In the long run, though, everybody gets chosen about as often as they choose, so it shouldn’t (he said cautiously) raise any eyebrows on the ratings committee.

Bill Smythe

8

It might have to be rated as a match (meaning that there are limits to how much a player’s rating can go up or down), but I don’t see any reason why it would be unratable. And if it is submitted online, like about 95% of events are these days, I’m not sure that the automated software would even require the match flag to be set.

9

I’m sure it wouldn’t, since the report would be submitted as a Swiss and there would be no repeat pairings. In other words, an organizer today could still get away with that format, in all probability.

But I think the ratings committee might raise an eyebrow or two if they found out. The “choose your opponent” concept can be carried too far. IMHO, though, with the algorithm I described above, there shouldn’t be any reasonable objection, for one thing because no player has the right to refuse to be chosen.

Bill Smythe

10

I don’t see anything inherently wrong with that format. A standard swiss is more about determining an overall winner for a tournament. If that isn’t the concern, then does it really matter who played whom as long as they were different opponents - at least for rating purposes anyway?

11

That’s the point I made upthread, statistically the ratings formulas DO NOT CARE who you played or why! The primary reason the match rules exist is that playing many games against one opponent runs the risk of violating the ‘law of large numbers’ that makes statistical predictions possible.

Another possible exception is if a really high rated player plays only players rated much lower. In such a case it is possible for the high rated player to slowly inflate his rating with little risk of actually losing or drawing a game against some fish. There have been a few cases of this.

I have suggested one way around this, which would be to require that someone crossing a 200 point boundary have recently played at least one opponent rated no lower than the 200 point group he is currently in. In other words, an A player looking to make Expert has to have recently played another A player and an Expert looking to make Master has to have recently played another Expert.

But like ratings floors this is probably more of a political issue, not a mathematical one (especially deciding what ‘recently’ means.)

12

The automatic system will flag any section as a match where the opponents play each other two or more times and no one else. An interesting side effect of this is a double blitz. We had a case at the last National K-12 where because of a fluke circumstance a double-blitz game had to be put into an extra rated game section. The system flagged it as match and refused to rate it because one of the opponents was unrated and it was technically two games against the same opponent.

13

One time I had an extra-rated games section (for cross-sectional pairings of the bye players) and it ended up having only two players in it having played one game. Because of the section difference they fell outside the 400 points and the section was flagged as a match violation and unable to be rated. I deleted the section, put both players and the game result in another section, and then it rated with no problems.

14

Don’t floating point ratings basically solve this (or at least make it take 100 times as long)?

Alex Relyea

15

The expectations were that going to floating point ratings would help, because there were cases where someone earned only a fraction of a rating point, and it got rounded up to a full point.

Whether in practice it has helped much with that problem has not yet been studied.

The office can override the match limitations warnings upon request (and with an adequate explanation of the cause, of course.)

16

Highly unlikely, in the “choose your opponent” format I outlined above. The high rated player will have his choice in round 1, but in round 2 any player with a rating of X will be a prime target for being chosen by players in about the X-minus-200 through X-minus-100 range.

Bill Smythe

17

The main point, though, is that the ratings system doesn’t care about pairings, so a ‘pick your opponent’ tournament is ratable, as is a ladder. The match limitations exist for mathematical purity, which is not an issue if in a pick-em tournament players aren’t allowed to pick an opponent they’ve already played unless they’ve already played everyone (in their set of choices.) And the ‘play only much lower rated’ situation is clearly an attempt to inflate one’s rating, which is also undesirable.

18

Depends on the players involved. I can easily see a number of players in a club wanting to have a shot at the top gun while the second and third highest would prefer playing weaker players and would thus avoid choosing the top gun. (this assumes the top gun is fine letting other people make the choice)

19

In any tournament I’ve ever played in, there are always lots of players who want to get a shot at the top players.

Of course, not every player feels this way, but enough of them do so that a strong player who wants to play exclusively against weakies will be thwarted in his attempt.

It’ll also never work. The floating point ratings will take care of that. Plus, in all likelihood, the strong player will suffer a draw or two along the way, and his rating will go down.

Bill Smythe

20

I would have said “It’ll never work” back in the days of integer-only post-event ratings, but then we found a number of players who did it quite successfully, one of them moved up 2 full classes that way, seldom (if ever) playing anyone rated within about 250 points of himself.

But as I said upthread, I’m not sure floating point ratings really fixed the problem, because in many cases the players were already earning more than 1 point per event before rounding. It may have slowed them down a bit, though. I haven’t looked for any examples since we went to FP ratings.