Time and/or Draw odds in rated events

Running Chess Tournaments
1

Fellow TD’s

I’m planning on hosting an event with the following format: 4SS G/90 for white, G/70 for black; black has draw odds (meaning that if the game is drawn, black is then awarded the point).

After I finished putting together the TLA, a couple of thoughts occurred to me: 1) can a game ending in a draw be reported as win even if it is advertised in advance, and 2) Can a time control legally allow more time for one side than the other?

I’m thinking that I may need to use the win for pairing and prize purposes, but after the tournament, tweak the results (Swiss Sys) such that drawn games are actually reported as drawn for rating purposes.

I’m hoping the above format wouldn’t make the event unratable. I’m often looking for new and fun formats to try, and I got this idea from a World Open tiebreaking method (7/5 blitz with draw odds for black).

I combed the rule book and was unable to locate anything to the effect that this would be in violation of USCF policy, but I’m not certain, so any understanding that you all could provide me with would be appreciated.

2

Time or piece odds games cannot be USCF rated.

The ratings system does not CARE why players are facing each other (except for the match rule restrictions), so as long as a draw is reported as a draw for rating purposes it makes no difference how it is treated for pairing purposes.

3

Very Good, Thanks Mike.

4

I brought up this topic some time ago under the heading of:

Games with different time controls for the players

and the consensus was that if it were advertised in all advance publicity, that such a tourney could be run, and be rated–were we all mistaken, then, or does the concept of draw odds add another factor into the equation?

5

I’m not sure what thread you are referrring to. I’m fairly sure that those in a position to cite official USCF policy (myself, Tim Just, Ken Sloan, etc.) have been very consistent on this point:

Both players must start with all pieces (in the standard position, Chess960 is NOT CURRENTLY USCF RATABLE) and with identical clock settings, anything other than that is NOT USCF RATABLE.

(Opening theme tournaments are USCF ratable, but what those are is an agreement by all participants to make the designated opening moves.)

Draw odds (as a pairing or prize allocation device) have no impact on the ratability of any specific game, and so they are permitted, as would other pairing systems: Swiss, Dutch, round robins, ladders, knockout events, random pairings, etc.

Why are piece odds or time odds games not USCF ratable? Because there is no clear-cut formula to compute the expected results of piece or time odds games.

Let’s take the simplest example, games between two identically rated players. Each player’s chance of winning is .50, so if the game results in a draw neither player’s rating will change.

Now, suppose White gives up a pawn or 10 minutes of time (in Game/30.)

What does that do to the expected result? Is Black now more likely to win, and how much more likely? And if this game winds up a draw, is Black now expected to lose some ratings points and White gain points?

It is worth noting that color is not taken into account by the USCF rating system, and I think Elo’s book covered this issue too. There are not any reliable statistics on White’s winning edge (and any such statistics would likely be dependent upon the ratings of the players), and the assumption is that over a large enough number of games that color assignment should balance out and thus cancel out whatever slight advantage White has.

6

Just for reference, nolan: the thread I was talking about is

USChess.org Forum Index → Tournament Organization → Games with different time controls for the players

And if I had gotten the wrong impression about the rateability of events with different time settings, or with other factors, then that is my fault–but with the information that I’d had, I got the impression through the quoted thread that if it were advertised sufficiently before the event, and at the event itself, then it would be rateable.

Extends wrist to be slapped, if need be :slight_smile:

7

The only response in that thread from anyone who could be considered a USCF ‘authority’ on the subject was the final one in that thread, from Ernie Schlich, former USCF Technical Director and an NTD.

His response was that games where both players do not have the same amount of time (aka time odds games) are NOT USCF RATABLE.

Hopefully my previous answer not only gives the proper USCF policy but explains WHY that policy is necessary under the USCF Ratings System.

This is one of those questions that comes up periodically and it belongs in a FAQ on the USCF Rating System.

8

.
To my understanding these quoted claims below seem blatently false. I would be curious and interested to read if anyone can post an explanation of how they could possibly be true. Thanks.

I cannot understand any basis for this claim Exactly one player will have the White pieces, and huge databases show White wins more often than does Black.
The exact modern ratio of WhiteWins -dividedBy- BlackWins will always be in slight flux. But if plain counts show the actual ratio is closer to say 1.17 than to 1.00, then using 1.00 is creating more error in the ratings measurement than would 1.17.

Huge databases that have accumulated millions of games from a century of chess do not meet the definition of ‘reliable’? Are we rejecting all things that are imperfect? That will be a big list.

I believe this is an incorrect statement, because in its grammatical wording it implicitly switches subjects (from people’s color assignments to the colors themselves).
“Balance out” what? Color balancing evens things out per player, not per color. White will still win more often (unless purposely biased/illegal color assignments are made so that the stronger players get Black in most 2-player pairings).

9

By ‘reliable’ I mean statistics that can be used for computing expected winning probabilities for the purposes of computing ratings, for identically rated players and for players who are not identically rated.

If a 2750 player plays a 2700 player, what’s his winning advantage with White versus the 2700’s winning advantage with White?

Yes, in games between GMs there are some statistics regarding White’s advantage, but are you willing to claim that same edge exists between A players, or between C players or between players rated 500?

FIDE doesn’t treat color differently for rating purposes, and I’m not aware of any national rating system that does, mostly because there are no agreed upon standards for how to adjust the expected results probabilities.

And the ratings system is designed with aggregate results in mind, at which point the number of games with White and with Black should be roughly even, along with the .50 expected performance for players with identical strength.

10

Gentlemen,

I was just looking at tournaments in the area and came across this TLA: Dec. 30 16th HOLIDAY OPEN GPP: 10 Enhanced Oklahoma 4-SS, G/80 white, G/85 black. Holiday Inn, 2515 W. 6th Street (Hwy-51), Stillwater, OK 1-405-372-0800. Free Parking. $$200 1st $$100 2nd. Other $ per Entries. Reg: 9-9:45am. Rds: 10-1-4-7. EF: $20 OGP #4. OCF req $10. Jim Berry P O Box 351 Stillwater, Ok 74076. 405-762-1649. jaberrycg@aol.com. CMV, LS, W.

I sent Mr. Berry an email asking him if he was aware that time odds events could not be rated. If this is the same person that I see listed as the only Jim Berry in the USCF member database in Oklahoma, then he is not a certified TD (or member for that matter), so I’m not sure that he’s even expecting this event to be rated. It’s just interesting to me that the USCF would list the ad for him. I guess it’s a good business move on both sides.

11

Fair point.
Yet I might argue the class A B C stats are irrelevant. Grandmaster play is closer to the ultimate “truth” about chess, so using stats from A B C would only add noise. The truth is as proper to apply to A B C as it is to masters.

It is a little unfortunate that the win-loss per-player per-game stats kept by the USCF today lack color info. Otherwise we could now quickly answer your interesting class A B C question. 30 years from now automated digitization should make it simple to capture this color info.

I get it now, thanks.
Still, using a factor slightly larger than 1.0 would introduce less noise, less noise that needed to be overcome with aggregation.

I would not suggest the USCF consider leading the way on this issue, no matter how strongly I believed in it (which is not very strong anyway).

12

Themed Opening: 1. e4 e5 2. Qg4 Qg5, now begin. Technically this is not “piece odds”.

The perspective that in an opening theme game the participants “make the designated opening moves” is a little arbitrary. There is precious little difference between making required initial moves versus starting from an alternate initial setup. Perhaps the concept could be:
“Play must begin from any position legally reachable from the one traditional initial setup”.

{Chess “Opening Theme S#549” , RNBB-KNQR}

  1. e3 e6 2. Qf3 Qf6 3. Be2 Be7 4. Bd1 Bd8
  2. Ne2 Ne7 6. Ng3 Ng6 7. Qe2 Qe7 8. Qf1 Qf8
  3. Qg1 Qg8 10. Nf1 Nf8 11. begin-any

BAP is more than a “pairing system”. It creates a conflict of interest that, in theory, certainly can “impact the reliability” of a specific game.
Tell me I get $0 if I draw with White, but I get $1,000,000 if I win. I feel it is safe to say that the Win-Lose-Draw counts from such games would be different than from regular games. Indeed I believe this is exactly the purpose of BAP (along with elimination of quickie draws).
Not that I mind.

So long as any reasonable-yet-unconventional chess games represent a tiny percentage of all rated games, the games should be rate-able.

13

I doubt you will be able to sell that concept to the USCF Ratings Committee, the Executive Board or the Delegates.

You’d probably have a better chance of selling them on rating Chess 960, and I consider that highly unlikely at present, mostly because there aren’t enough Chess 960 events to support it as a separate rating system and not enough generally accepted evidence as to whether chess skills would be uniformly transferrable to Chess 960 performance to include those games with regular rated events.

As to your thematic tournament, would YOU pay $20 to enter it?

14

In Michigan we have an annual event called the Bottom Half Class Championship. When this event started about 8 years ago, Bob Ciaffone asked the USCF if he could rate the event if some games started with black having extra time (not much, maybe 10-20 min more) and they agreed.

As far as I know, this tournament no longer does this (I think all players now have equal time in all games), but it did for 3-4 years.

Ben Finegold

15

I don’t recall offering an opinion on time-odds games. Nor do I recall any official pronouncements on the topic. Personally, I don’t see a problem with time odds. To my mind, time odds are LESS unusual than theme events. I agree that shuffle-chess should not be rated.

16

I thought you were one of those who had said that both players have to have identical time and material (except in the case of thematic events), Ken. If not, I stand corrected.

I looked through the current rulebook last night, and couldn’t find anything on the issue of odds games or thematic tournaments.

This is one of those areas where I don’t really care what the rules are, as long as they are clearly stated and consistently applied. Perhaps the Ratings Committee or Rules Committee would be willing to make a recommendation to the Delegates on these issues?

17

So then is the current understanding that USCF policy allows 1) time odds to be permitted if advertised? and 2) draw odds to be permitted for pairing and prize fund calculation purposes (if advertised), but reported as a draw for rating purposes? Id be interested in hearing input from someone on the TDCC chair (Tim if you’re here, since we’ve met.)

18

Actually this is not a TDCC question but a rules question. So I will say I believe (2) is correct, if advertised. It may fudge with the rules regarding pairings and prizes but it is advertised! I am not sure about (1). It seems to fudge with the basic assumptions (standards) on how ratings are calculated based on statistics, predictions, … Just because someone may have done it does not mean it is permitted. The office staff is human and things do slip by them.

Tim

19

I agree with Tim. Draw odds, as long as they are reported as draws for rating purposes, are USCF ratable events.

And ANY deviation from the expected conditions (in this case swiss pairing rules) should be advertised in advance.

20

Thanks Gentlemen,
I do appreciate your time and input.