PDA

View Full Version : Game order within a period...

The_Wise_Man
04-05-2005, 11:53 PM
Just a quick question....

How are games processed in the system??? is each game mutually exclusive.... Otherwise you get more/less rating points from a player depending on your run?

For example say a 1400 player has had a good run and is about to gain 80 points (from previous games within the same period), the player meets a 1450 player and draws, does the player receive or lose points under this scenario?

Wise

Bill Gletsos
05-05-2005, 12:43 PM
Just a quick question....

How are games processed in the system??? is each game mutually exclusive.... Otherwise you get more/less rating points from a player depending on your run?

For example say a 1400 player has had a good run and is about to gain 80 points (from previous games within the same period), the player meets a 1450 player and draws, does the player receive or lose points under this scenario?

WiseGame order is totally irrelevant.
All games in a rating period are processed and an intermediate rating for all players is generated. For the calculation of the player's rating his rating at the start of the period is still used but the opponent's intermediate rating is used as the opponent's rating rather than their rating at the start of the rating period. Testing shows that it is indeed better to do this as it leads to better predictive accuracy. It should be noted that the USCF uses this intermediate rating idea in their rating calculations.

Garvinator
05-05-2005, 01:03 PM
I have wondered, what happens with games played against unrated players?

Bill Gletsos
05-05-2005, 01:27 PM
I have wondered, what happens with games played against unrated players?I have answered that numerous times previously, the most recent being here http://www.chesschat.org/showthread.php?t=1554

Bereaved
18-11-2005, 08:18 PM
Quote:
Originally Posted by The_Wise_Man
Just a quick question....

How are games processed in the system??? is each game mutually exclusive.... Otherwise you get more/less rating points from a player depending on your run?

For example say a 1400 player has had a good run and is about to gain 80 points (from previous games within the same period), the player meets a 1450 player and draws, does the player receive or lose points under this scenario?

Wise
Quote:
Originally Posted by Bill Gletsos
Game order is totally irrelevant.
All games in a rating period are processed and an intermediate rating for all players is generated. For the calculation of the player's rating his rating at the start of the period is still used but the opponent's intermediate rating is used as the opponent's rating rather than their rating at the start of the rating period. Testing shows that it is indeed better to do this as it leads to better predictive accuracy. It should be noted that the USCF uses this intermediate rating idea in their rating calculations.

Hello, Bill
Given the statement of the games order being unnecessary, am I correct in remembering that the tournaments are rated in the order they end, in sequence? Or is it the order that they start?

The reason that I ask is that even given the games being rated based on an individual's initial rating vs their opponents' intermediate ratings, the person one played at the start of a rating period may have established an intermediate rating at marked discrepancy to the rating their opponent perceived them to be when they played them by the end of that list.

How if at all is this adapted to make an accurate rating adjustment?

Also, and forgive me if this has been asked before, so redirect me if you want to the appropriate post, which rating is used when a game which was played in an event which commenced prior to September 1, 2005, but the given event ended after this date?

Thanks in advance for your help,

Take care and God bless, Macavity

Kevin Bonham
18-11-2005, 10:42 PM
macavity, "intermediate" doesn't mean halfway through the ratings period - it means halfway through the calculation process. The order in which your games are played and/or submitted, if they are part of the same ratings period, is utterly irrelevant.

Bereaved
18-11-2005, 10:55 PM
Hi Kevin,

I will wait for Bill's answer as you are not answering my question.

Read my post again, if you like. I refer to the fact of how my opponent's intermediate rating in the course of a rating period total may become (potentially) dramatically less/more than I thought it was when I played them.

I refer to Bill's statement that this ( the intermediate rating) is the figure which ultimately is used to adjust my previously published rating

Take care and God Bless, Macavity

Bill Gletsos
18-11-2005, 11:13 PM
Hello, Bill
Given the statement of the games order being unnecessary, am I correct in remembering that the tournaments are rated in the order they end, in sequence? Or is it the order that they start?Tournament order is unimportant, like game order is unimportant.

The reason that I ask is that even given the games being rated based on an individual's initial rating vs their opponents' intermediate ratings, the person one played at the start of a rating period may have established an intermediate rating at marked discrepancy to the rating their opponent perceived them to be when they played them by the end of that list.Quite true but ones perception of an opponents strength doesnt always mean much.

How if at all is this adapted to make an accurate rating adjustment?As I said the game order/tournament order is unimportant.
e.g. Player A is rated 1800 at the start of the period. A few days later he plays a player rated 1800 and wins. Throughout the period the player plays many games. At the end of the period he plays another 1800 player and wins. Over the entire period he performs around the 2100 level.
What strength is player A when he beats the first 1800 and what strength is he when he beats the last 1800.
Should he be treated as
a) 1800 for both
b) 1800 for the first and 2050+ for the last.
c) some intermediate value for both between 1800 and 2100.

a) is the FIDE method, b) game by game in chronological order like online servers, c) is the USCF and ACF method

The answer is easy.
It is the one the leads to better predictive accuracy.

Testing shows that the one that does that the best is c).

Also, and forgive me if this has been asked before, so redirect me if you want to the appropriate post, which rating is used when a game which was played in an event which commenced prior to September 1, 2005, but the given event ended after this date?I have answered this previously.
As has always been the practice the ratings used are the ratings at the time the event is submitted for rating.

Kevin Bonham
18-11-2005, 11:24 PM
Read my post again, if you like. I refer to the fact of how my opponent's intermediate rating in the course of a rating period total may become (potentially) dramatically less/more than I thought it was when I played them.

OK, I got that ... finally ... your first sentence confused me because it seemed like you were expressing a view that order of ratings somehow still mattered.

The impact of playing an unrated player on your rating is nothing like as great as playing a reliably-rated player of the same rating, so even if there is a dramatic difference between how good you think the unrated player is and the rating they get, it won't affect your rating all that much. Indeed a player with a known rating having an unusually strong/weak ratings period could have a similar impact.