provisional players over 1200 in RBO?

Running Chess Tournaments
1

We’re having a series of Rated Beginners’ Open events (open to players under 1200). These are G/30 d5 dual-rated events. Entry fees are low, as are prizes (no cash prizes) clock & chess book for first, two books for second etc. We’ve been getting 20-32 players for these events, which is nice!

What to do with provisional players? I allowed a player with no regular rating & a provisional quick rating of 1375/10 (not the real #s but close enough) into our dual rated event. This seemed clearly OK to me.

A player emails me today: “I’m 1325/9 (regular)” and I said yes again. This one I was not so sure about.

As a practical matter, if organizers are being consistent & fair, no one will complain. But what is best practice? And should one pair 1325/9 as an 1125 or a 1325? And how about a 1325/20?

2

No difference. You can increase a player’s published rating if you like and have reason, but you can’t decrease it.

Alex Relyea

3

Established (rating based on 26 or more games) vs provisional (rating based on 25 or fewer games) is not supposed to make a difference. What should matter is official (published in the appropriate monthly “supplement”) vs unofficial (appearing only in crosstable(s) for event(s) that came after the appropriate monthly “supplement”).

If the tournament announcement says “open to players rated under 1200 or unrated”, then a player with an official (published) rating of 1200 or over, whether established or provisional, should not be allowed to enter.

If a player has no regular rating, you may use the player’s quick rating to determine eligibility.

Whatever rating you use to establish eligibility, you should use that same rating to determine pairings. I’m pretty sure all pairing programs allow the TD to override whatever rating the program may have dug up from its database. The rating you specify will appear on the pairing sheets and wall charts, and will be used for pairings.

The rating you use in the tournament report to US Chess makes no difference. US Chess will use the latest rating from its files (whether official or unofficial) to calculate the player’s new rating and the player’s opponents’ new ratings.

Read rule 28C (pages 118-119) in the US Chess Official Rules of Chess, 6th edition, for more details.

If you look up any player in MSA, read the General tab (not the Tournament History tab) for further information about official vs unofficial ratings. The word “supplement” is a throwback to the days of paper rating supplements that were issued monthly as updates to the annual rating list.

Bill Smythe

4

Any player who has played at least four rated games, either Quick or Regular rated, and has a rating above 1199 should not be allowed to enter.

5

I’d agree except for the caveat that if it is a regular-rated tournament then a regular rating of 1150 would make the player eligible even if the player also has a 1220 quick rating. If a TD uses and pairs people using the higher of multiple ratings then the rating used for pairing purposes as to be under 1200.

6

Thanks for the clarification. Would the following be acceptable?

  • Use regular ratings (including provisional) for both eligibility & pairing purposes. Ignore quick ratings for players with regular ratings when determining eligibility.

  • If no regular rating, then use established quick for both eligibility & pairing purposes.

  • If no regular rating and no established quick rating, then use provisional quick rating based on N games to determine eligibility & pairing: (Provisional Quick Rating) - 10*(26-N) = Eligibility & Pairing Rating.

Our goal for these events is to create something like a stop on the vaudeville circuit: a safe place to have a bad performance & move on… I would like to have the most inclusive criteria possible while treating all the beginners fairly.

Some of these adult beginners (the ones with established ratings in the 900s-1100s) are pretty good players already! I watch them play, and I’m not sure why they aren’t already 1300s

7

I think that if a player does not have a published regular rating, the organizer/TD is PERMITTED to use a quick rating for eligibility and pairing purposes, but not REQUIRED to do so. Personally, I would probably consider a quick or blitz rating for pairing purposes but not for eligibility purposes.

8

The OP’s question concerned a player who had one rating, but not both, and that was what I referring to. I guess I should have phrased my reply more precisely, but I was referring to a player who had regular rating but not a quick rating, or a quick rating but not a regular rating. If the player has both, then we are in agreement.

9
  • That’ll work, and it’s the book answer anyway.
  • Ditto. Certainly quick ratings can be used for players without regular ratings.
  • In other words, relax (increase) the 1200 upper limit by 10 points for each game the player’s number of quick games falls short of 26. Perhaps reasonable, if nobody squawks. And, if you use your formula for “Eligibility & Pairing Rating” on the pairing sheets and wall charts too, then all the ratings will appear to be below 1200 anyway, so in all probability nobody will notice – unless they start looking up ratings themselves while the event is going on. It seems unlikely, in an event of this sort, that you’d get caught, or that anybody would care.

Bill Smythe

10

Thanks again…I emailed the 1325/10 player back & told him my original answer was wrong. Fortunately, he’ll be entering the other event.

We will however be more permissive with provisional quick ratings.

11

Well, OK, but this seems barely justified (or perhaps not quite justified) under 28C, 28D, etc. You have to do a lot of rule-bending, almost to the breaking point.

In speaking about “players who are known to have ratings … of other types, such as … USCF Quick …”, 28D1 says “it is recommended that such players not be considered unrated and that their [ Quick ] ratings be used …”.

Since that’s just a recommendation, you could ignore it and consider the player unrated, in which case he is eligible for the RBO since that section is open to under 1200 and unrated. Or is it? “And unrated” does not appear in the tournament announcement on your website. OTOH, the next sentence says “You do NOT need a US Chess rating to play in this event!”

OK, so now the player is in the RBO, but how do you assign a rating? 28D, Players without USCF ratings, tells you what such players are eligible for “unless alternate procedures are used to assign ratings (28E)”, but 28E is about “Assigned ratings for rated players” (underscore mine). What gives?

The entirety of rule 28, or at least 28C through 28E, is a hodge-podge of unclarity and contradictions, and most of it dates back to the pre-electronic days of pairing cards, printed rating supplements, etc. So what is an organizer supposed to do?

In any case, if this player with no regular rating and a provisional quick rating comes out of this event with a post-event regular rating of 1200 or over, I fervently hope you will strongly discourage this player from entering future RBOs, even if that 1200+ rating isn’t official yet because of the timing of the rating “supplements”.

Bill Smythe

12

Follow-up: If the player with no regular rating and a provisional quick rating is who I think it is, your problems are solved regarding this player in your future tournaments. Both of his post-event ratings (regular and quick) are now below 1200.

Well, almost solved. Those ratings do not become official until August 1. But because he has no official regular rating until then, you could still use his unofficial regular rating, I suppose.

Bill Smythe

13

“Barely Legal RBOs”: catchy title, but given recent events in Illinois, we will pass.

14

Bill,

I’ve wondered if some of the Rulebook information on ratings is “correct” – and would also point out that its not clear to me that the definition you give of “provisional” is currently correct.

According to The US Chess Rating System authored by Glickman and Doan and dated April 24, 2017, found at glicko.net/ratings/rating.system.pdf the definition of the Special Rating Algorithm (“this used to be called the “provisional” rating algorithm” – according to the paper), the definition is whether or not the person has greater than, or fewer than, 8 games, so long as they haven’t had all wins or all losses.

It’s true that the first part of the paper defines an established rating as more than 25 games – although it doesn’t appear that any aspect of the rating formula is dependent upon that. It also doesn’t define provisional as being fewer than 26 games.

My sense is that ratings are no longer provisional in the same sense in the past, but they are established at 25 games and the “effective number of games” can never be greater than 50.

So, in the sense of 28C or the questions asked by Bill Brock, I don’t really know what “provisional” means.

15

There’s no longer any connection between the rating formula used and the provisional/established boundary. The 25 is actually the number which really should have been used in the old K=32 regime because 25 games is the point at which the linear method used in the old provisional ratings uses similar weights on old vs new results as K=32. However, we try to get ratings off the linear system (with all the tweaks) as soon as reasonable, thus after 8 games for most players unless they have all wins (as if!) or losses (unfortunately all too common).

16

Is there still less weight given to the first 10 games? It use to be that since a new player may not be use to rated tournament chess [clocks, time controls, score keeping, etc] that the first 10 games were not as heavily weights as the next 15 in getting to an established rating [so I was told way back when].

Larry S. Cohen

17

The P for provisional is usually removed after the 26th game is rated, but I’ve seen P46 if all 46 games were losses.
Granted, that provisional rating was not over 1200.

18

There used to be a different formula used for the first 20 or 25 games (in effect averaging the win+400/loss-400 point performance rating formula over those games), but that flawed approach was abandoned decades ago.

These days, there’s a special formula used for players who have 8 or fewer games, though it has some characteristics of the old performance rating formula. It’s not so much that it weights those games differently as it treats the player as having a less accurate rating, thus one more subject to changes.

For a player who has 9 or more games at the start of an event, the same ratings formula applies regardless of whether a rating is provisional or established (including bonus points), except that players with fewer than 26 games do not get peak rating based or money prize floors.

For the gory mathematical details, see glicko.net/ratings/rating.system.pdf

Players also remain provisionally rated if they have all wins or all losses. I think the record is 53 games lost in a row before the player finally won a game.

It is probably true that the ratings formula has gotten more accurate mathematically but is harder to explain to chessplayers.

19

Doesn’t the K-factor vary based on the number of previous games? Perhaps it starts out larger, then gradually diminishes until 50 games (or some such number) is reached?

Bill Smythe

20

The ‘effective number of games’ is a function of the player’s rating, as lower ratings are considered less reliable. (I think of it as a mapping of reliability into the number of games played, the higher the rating, the higher the effective number of games.)

Although the number of games played previously is used as an upper limit on ‘effective games’, neither the effective games formula nor the regular/special formulas directly use the number of games that the player has played in previous events, though the number of games previously played is used to determine whether the regular or special formula is used.