Rating Questions

  1. I understand that a provisional rating requires four rated games. Therefore, an unrated player plays in a tournament of three rounds, they would not have a provisional rating. However, if they play in another tournament of 3 rounds, would this still result in no provisional rating being provided? Or, would the rating be based upon the cumulative performance of the six rounds?

  2. In that second tournament, how would the player be paired? I assume that they would be “unrated”, but should the TD use the performance in the previous tournament of three games to assign a rating for pairing (28D4)?

  3. If a player has a provisional or official Quick Chess USCF rating but does not have a Regular USCF rating, my understanding is that the Quick Chess rating is to be used at a Regular time control tournament (28D1a). I’m I correct? This also works in reverse for those who have Regular ratings but not a Quick rating, right?

Thanks in advance.

An “official” rating requires at least four rated games. Am unrated player who plays in a tournament of 1-3 rounds gets a provisional rating but that rating never becomes “official”.

For example, I have a player today with a rating based on 3. He’s playing one game and going home. He’ll have a published rating in the May supplement. Mike Nolan can correct me if I misunderstand, but until a player has played eight games, or has a result that is not all wins or all losses, his rating is always based on his total performance.

Ratings based on fewer than four games are so unreliable that I would never use them for pairings and prizes. If a player has an unpublished rating based on four or more games, then I’d use 28D4.

This falls into the “better than nothing” category. It’s extremely unsatisfactory, however, but unrateds are also unsatisfactory. There is really no good way to handle them, unless we are willing to adopt FIDE’s approach, and I see that as impossible to sell to the delegates.

Alex Relyea

Until a player’s pre-game rating is based on 9 or more games (and not all wins or all losses) the special ratings formula is used. It’s similar to the old ‘wins + 400 and losses - 400’ formula, but not quite identical to it. After the player has 9 or more games (and not all wins or all losses) the regular ratings formula is used.

As to when the rating is published, it takes at least 4 games, but the 4 games don’t have to all be in one event.

Note that it IS possible to earn norms even if one has fewer than 9 rated games, because norms are based on how one performed compared to the expected performance at that norm level, not on the player’s current rating.

It’s cumulative.

The player is still considered unrated, and should therefore be considered ineligible to play in a rated (as opposed to an open) section. But you have to assign them some sort of rating for pairing purposes, so the question is whether assigning them a provisional rating based on only 3 games will do a better or worse job of placing them than the rating you would assign them if they had no rating at all.

In quads, for example, I would consider mixed results to be an indication that the player was probably well placed. If, on the other hand, an unrated player who wasn’t in the top quad won all 3 of his games, I would suspect that they were underplaced (i.e., they probably belonged in a higher quad), and if an unrated player who wasn’t in the bottom quad lost all 3 of their games, I would suspect that they were overplaced (i.e., they probably belonged in a lower quad). Based on the dozens of tournaments I’ve directed over the years, it has appeared that assigning an unrated player their 3-game provisional rating nearly always does a better job of placing them than using the rating I would have given to a player who had no rating at all.

Yes.

Bob

Fantastic reply, Bob. Thanks!

My next question is how to enter these ratings into WinTD.

There are two scenarios…the player who has only the three games or the scholastic player who has no rating. From what everyone has said, the 3 games rating is better than nothing, so I’ll use that.

For the scholastic player, if I understand the previous posts correctly, you take the age of the player (including the decimal) and multiply by 50 to provide an initial estimated rating.

Do I enter these “ratings” as local ratings or as the USCF ratings on WinTD in order to pair and ultimately develop their rating?

Thanks in advance!

Depends upon how you plan to use them. If you want to take unofficial ratings and use them for pairing purposes just put them in directly into the “Rating” field. You can also put them into the “Local” field if you have the section set for Pairing Numbers By—Main Rating, Local if Unrated (which is the default so it’s probably what you have set anyway). The two end up producing the same pairings. Since they are US Chess ratings (just not yet official), I would put them in as just the Rating if I’m using them that way.

Once you run some actual U.S. Chess rated events, for reporting purposes (to U.S. Chess) it won’t matter what, if anything, you put in the rating field, since the U.S. Chess rating software looks everybody up in their database anyway. If, however, you put in a rating that differs by hundreds of points from what U.S. Chess has, their reporting software might flag it as a warning and ask you to check it.

For pairing purposes, you can use whatever ratings you feel are appropriate, for whatever reason. Then, after all the games in all the rounds are complete, and before you prepare the rating report, you can go back and change the ratings to something close to what U.S. Chess has, to avoid warning messages. At this point you might also want to change any players to Unrated if they have no prior U.S. Chess rating history at all (again, this would be to avoid warning messages).

Once you have run a few U.S. Chess rated events and most of your players have actual ratings, you might find it no longer necessary to maintain your own informal ad hoc database of ratings for unrated players.

Bill Smythe

Not if they’re lower than the published rating.

Mike Nolan would have to clarify that, but I wouldn’t think showing a rating for an (officially) unrated player would trigger a warning. The point is to catch what may be ID typos—for instance, before there were checks (because they’re couldn’t be, as the upload format didn’t include the TD’s value of the rating), an IM’s son’s ID was mistakenly put in for the IM himself, so an 800 rated ID apparently popped for a 2600 performance.

I wouldn’t think showing any rating for a player would trigger a warning, regardless of how out of whack that rating would be. The upload pulls in the current rating for a player based on the ID number. There are warnings for weird performances (that IM’s son would now trigger a warning to be answered with the different name (and the 8xx rating that was retrieved even if the player was paired as a 2600 in the rating field) being red flags to let the TD know that the ID number is probably wrong.

PS I generally print the final wall chart in score order and then verify all of the names with the upload. That has helped catch typos in the ID numbers when the names don’t match up.

The event validation software doesn’t currently look at what rating was used for pairing purposes in an event, because the original upload format didn’t even supply that information.

You will get a warning if someone’s performance appears to be way above or way below the pre-event rating. This happens a lot with higher rated players who are having a bad day, lose a game or two, and drop out of the event with a zero score. You will sometimes see it with younger players who have just jumped up a quantum level in their chess ability.

But the main reason we do this is that it helps find ID errors. (The stereotypical example of this was a New Jersey master whose ID was used in an elementary school event in California, he was not happy when he lost several dozen points because he supposedly lost several games to players rated under 1000.)

I see this sometimes when someone enters an inappropriate section (or tournament). Someone is flagged for a low performance rating when he wins all his games, or a high one when he loses all his games.

Alex Relyea