PDA

View Full Version : 10. 2012-S Improved Bonus Point Formula



Lyle Craver
03-31-2012, 11:21 PM
Moved by Paul Leblanc
Seconded by Fred McKim

To replace the existing Bonus Point Formula (CFC Handbook Art. 714.d) with the following:

BONUS 1 = Rmax BONUS *a*Ke
BONUS 2 = b*RATING CHANGE BONUS*(Rnew – Rold – THRESHOLD)*Ke
TOTAL BONUS = BONUS 1 + BONUS 2

where:
a = 1 if the new rating is at an all time high, 0 otherwise;
b = 1 if Rnew > Rold + Threshold, 0 otherwise;

Threshold = RtgChangeThreshold*Ke*sqrt(n) where n is the number of games played; no bonus points are awarded if less than 4 games are played;

Ke is the ratio of the player's K factor to the K factor used for players rated under 2200; For the CFC rating system, K=32 for players under 2200 and K=16 for players at or over 2200;

Rnew is the post-event rating and Rold is the pre-event rating;
and RatingMaxBonus, RtgChangeBonus, and RtgChangeThreshold are constants with the following values:
RatingMaxBonus = 20; RtgChangeBonus = 1.75; RtgChangeThreshold = 13.

The numerical values in the bonus point equation may be adjusted from time to time by the Rating Auditor as deemed necessary and in consultation with the CFC Executive

Reason for Motion:

The proposed formula is based on detailed analysis of the existing tournament data base from 2006 to 2011 and is distilled from the examination of numerous options. It does a superior job of injecting rating points, i.e. when modelled, it generated ratings that more closely predicted actual game results than the existing system.

There were concerns that the existing formula awarded disproportionate bonus points to a small group of high rated players. The proposed formula improves this.

The rating system is inherently deflationary. Care was taken to select a formula that minimizes the effects of inflation and deflation.

The proposed formula mitigates the problem of under-rated juniors by awarding bonus points to players achieving new personal high ratings and by awarding a generous bonus to exceptional performances. These criteria are very common among juniors.

Detailed examination of the methodology leading to this proposal can be found at:

http://www.victoriachess.com/cfc/cfc_rating_doc.htm

Bob Gillanders
04-01-2012, 09:10 PM
I have read through all the material, and I think I understand the formulas. But maybe I got it wrong. Could you please give us a few hypothetical cases, and calculate the bonus for us?

For instance, player X has the following result.

Pre tournament rating = 1500
Post tournament rating = 1550
games played = 6
Highest ever rating = 1525
All opponents <2000

Thanks.

Egidijus Zeromskis
04-01-2012, 11:54 PM
A.
There are several math mistakes:
Rmax BONUS, RATING CHANGE BONUS - became RatingMaxBonus and RtgChangeBonus, etc.
That is not a scientific way to write down formulas and their explanations.

B.
Probably it is too late now but plots should include FIDE ratings.

C.
Still there are very huge jumps and downs.

D.
IMHO: No bonus points for players rated above 2199

Kevin Pacey
04-02-2012, 05:36 AM
D.
IMHO: Bonus points should be applied only to 2200 players.

Part of the rationale behind this motion is to combat the effects of underrated juniors, presumably meaning largely juniors rated below 2200.

To take away bonus points from sub-2200 players would be contrary to this part of the rationale, which is one part I certainly agree has merit.

Kevin Pacey
04-02-2012, 06:06 AM
D.
IMHO: Bonus points should be applied only to 2200 players.

IMHO, the CFC might consider eliminating bonus points for players rated above 2199, if we first pass the motion as is, and then find that rating inflation has resulted after some years. However I haven't a reason as yet to suspect that the modelling performed that helped produce this motion was incorrect.

Egidijus Zeromskis
04-02-2012, 08:37 AM
eliminating bonus points for players rated above 2199

That what I wanted to write... Corrected.

Paul Leblanc
04-02-2012, 09:24 AM
Fred McKim suggested in our discussions that players above 2400 not be eligible for bonus points. We decided to leave that discussion for this forum.
Egidijus, I'll make sure the variables are contracted/not contracted in the right places in the handbook/website. Is that your point about math errors?

Paul Leblanc
04-02-2012, 02:23 PM
Bob, here is the answer to your question:

a) as player X's rating is <2200, Ke = 1.

b) Rnew is at a new high so BONUS1 = 20.

b) Threshold = 13* sqrt(6) = 13.84: => BONUS2 = 1.75*(1550-1500-31.84) = 31.78

c) Total bonus = BONUS1+BONUS2 = 20+31.78 = 51.78; Rounded to the nearest integer, this is 52 points.

Note 1: in Roger's work, the bonus calculation was rounded to an integer at the end rather than at an intermediate stage (although intermediate ratings used in the calculation, e.g. Rnew, were rounded when they were calculated)

Note 2: It doesn't matter that player X's opponents were rated below 2000.

Egidijus Zeromskis
04-02-2012, 02:58 PM
b) Threshold = 13* sqrt(6) = 13.84

correct to 31.84

Bob Gillanders
04-02-2012, 03:05 PM
Bob, here is the answer to your question:

a) as player X's rating is <2200, Ke = 1.

b) Rnew is at a new high so BONUS1 = 20.

b) Threshold = 13* sqrt(6) = 31.84: => BONUS2 = 1.75*(1550-1500-31.84) = 31.78

c) Total bonus = BONUS1+BONUS2 = 20+31.78 = 51.78; Rounded to the nearest integer, this is 52 points.

Note 1: in Roger's work, the bonus calculation was rounded to an integer at the end rather than at an intermediate stage (although intermediate ratings used in the calculation, e.g. Rnew, were rounded when they were calculated)

Note 2: It doesn't matter that player X's opponents were rated below 2000.

Thanks Paul. Threshold = 31.84, right?

I tried out the new formula on a recent tournament of mine, the Burlington Club Championship. I did well. :)

Old rating = 1819
new rating = 1872
games = 5
performance rating = 2092

No bonus points awarded under current system. :(

New proposed bonus plan
B1 = 0
Threshold = 29
B2 = 1.75 ( 1872-1819-29) = 42

Excellent. My new rating would have been 1914. :D

I like the new bonus system.

Lyle Craver
04-02-2012, 03:17 PM
The major cause of ratings deflation is when players leave the pool having gained several hundred rating points from where they entered.

For me my biggest memory is the tournament AFTER the best tournament of my life (where I scored 3.5/5 with two master wins (both with black!), one expert draw and an A player loss and a B player win - it was one of those double-sided Sicilians where both of you are balanced on the edge of the precipice and it's NOT going to be a draw) and reached my lifetime high of 1905 only to lose in consecutive rounds to two 1400 rated brothers (juniors) who went to 2100 and 1950 respectively before deciding they enjoyed playing soccer better.

Net gain for me after the two events was about 20 pts...both brothers contributed a lot to ratings deflation in Vancouver when they left!

I haven't finished my analysis of the numbers but like what I see so far.

Egidijus Zeromskis
04-02-2012, 03:28 PM
Old rating = 1819
new rating = 1872
games = 5

new rating would have been 1914. :D

I like the new bonus system.

100 points in one 5-game tournament :eek:

Egidijus Zeromskis
04-02-2012, 03:30 PM
Moving (looking for a seconder) to amend:

"Bonus points are not calculated for players rated above 2199"

Bob Gillanders
04-02-2012, 03:38 PM
100 points in one 5-game tournament :eek:

Well, it is only 95 points, but why not? I did win all my games.

Is not the objective of a bonus system to quickly increase those underrated juniors, like myself? :D:D:D

Lyle Craver
04-02-2012, 03:38 PM
if you mean to insert that into the motion I will second that - LC

Egidijus Zeromskis
04-02-2012, 03:59 PM
if you mean to insert that into the motion I will second that - LC

Thank you.

Egidijus Zeromskis
04-02-2012, 04:02 PM
Well, it is only 95 points, but why not? I did win all my games.

With four more such perfect performances you will be eligible for the Olympic team. Congratulations :D

Rob Clark
04-02-2012, 05:04 PM
Excellent, Bob has found the fountain of youth. The CFC's financial problems are over.

Paul Leblanc
04-02-2012, 06:15 PM
We didn't dare go to negative bonus points for exceptionally poor performance.
I think people will like the opportunity to benefit from outstanding results. The points they gain will eventually filter down to the up and coming juniors (I predict) but with less pain than we have now.

Lyle Craver
04-03-2012, 02:13 PM
I have CERTAINLY had tournament performances like that!

The serious answer though is that mathematically the primary reason for deflation is that players (not just juniors but they're the main problem since the ones who leave have typically not been around for more than 2 or 3 years as opposed to adults who may have been in the pool for 20+ years) leave the pool rated considerably higher than they entered it.

Very few leave the rating pool lower rated than when they started.

Kevin Pacey
04-03-2012, 02:57 PM
Moving (looking for a seconder) to amend:

"Bonus points are not calculated for players rated above 2199"

Now that Lyle has seconded your amendment above, Egidijus, would you be able to provide a rationale for it at this time?

I'd repeat myself, more or less, in saying I haven't got a reason to vote in favour of this amendment, as I haven't seen a reason not to trust the modeling behind the original motion. However Paul Leblanc suggested 2399 as an optional dividing line (for no longer awarding bonus points), so I wonder if there was a variant model done for that option as well.

The old CFC rating formula that was around in the 1980's (if not before) chose 2299 as the dividing line. That old formula held up well for many years, I'm told, until it was changed (but by then the number of CFC members may have become significantly smaller).

Paul Leblanc
04-03-2012, 03:56 PM
Kevin, I'll relay this question to Roger Patterson to see what he thinks. I don't think he modelled any upper limits to the bonus point system but we discussed it briefly when Fred McKim suggested 2400.

Paul Leblanc
04-03-2012, 05:25 PM
I've discussed this with Roger and I recommend against this amendment for the following reasons:
1. there is insufficient time to re-test our models and analyse the results within the timeframe for this meeting;
2. the levels being discussed (2200, 2300 and 2400) seem arbitrary and the logic leading to the motion has been pretty thin.

What I suggest instead is that the governors task me with examining whether a cap on the bonus system would be beneficial in achieving whatever aim you feel you are trying to achieve by capping the bonus system (less volatility at the top? perceived inflation among top players? reduce fluctuation of standings to make Olympiad selection easier? other reasons?)

If I am given a task along these lines, I can come up with some options for the next meeting. In the meantime, we can implement the new Bonus Point System and perhaps get some real time data on the impact it will have on strong players.

Kevin Pacey
04-03-2012, 06:23 PM
Hi Paul

At the moment I like the idea of passing your original motion (i.e. without the amendment) and seeing what happens.

As far as the Governors assigning you any task of the sort you suggested, I would have thought that generally (or in any aspect) monitoring the rating system throughout a given term is covered in the Rating Auditor's volunteer job description in the Handbook:

From Section 2:

RATING AUDITOR

10. The Rating Auditor shall supervise the operation of the rating system and shall deal with rating appeals.

In short, I think this means you get to decide, with the help of any advice you receive (from Governors or otherwise), whether the rating system needs to be changed by the Governors at any point in time.

Kevin Pacey
04-03-2012, 08:34 PM
Personally (especially if my interpretation of the Handbook above is correct), I think the Rating Auditor position should have been left as an Executive position. However, I wasn't around [edit: as a Governor] for the vote on that issue, I'm pretty sure.

Egidijus Zeromskis
04-04-2012, 08:38 AM
Egidijus, would you be able to provide a rationale for it at this time?

A.
'cause I compared rating plots at FIDE for the sampled players from http://www.victoriachess.com/cfc/cfc_rating_doc.htm
They better matched when the CFC system uses no bonus points. As the both ratings still are in use for as a selection criteria , imho, they should be equivalent without big differences.

B.
above 2199 (or >=2200) - a master level rating (even FIDE long time ago used as a floor for its rating). And the difference between classes/categories 2300/2400/2500 are significant. However, the bonus system may add ~100 points per tournament on regular bases (the Gillanders' case). In the Leblanc's case, the bonus was ~200 points, though after that the rating slowly goes down.

C.
IMHO :)

Gordon Ritchie
04-04-2012, 11:02 AM
I support Paul's suggestion.

Lyle Craver
04-04-2012, 04:09 PM
*** NOTICE TO GOVERNORS ***

Mr. Zeromskis has made an amendment to this motion that would delete all clauses relating to players rated over 2200 (seconded by myself for discussion).

This amendment is in a 'sticky' labelled CALL FOR VOTES and I urge each of you to respond.

The voting deadline is Thursday at 9pm ET.

Thanks,
Lyle Craver
Secretary, Chess Federation of Canada

Paul Leblanc
04-05-2012, 10:15 AM
The proposed amendment seems not to reflect Egidijus' proposal to remove 2200 players from the equations.

Michael von Keitz
04-05-2012, 10:32 AM
The proposed amendment seems not to reflect Egidijus' proposal to remove 2200 players from the equations.

It seems to have been correctly stated in the preamble to the poll. If the amendment passes, it will be understood as players rated above 2199, not above 2200.

Bob Gillanders
04-05-2012, 10:42 AM
It seems to have been correctly stated in the preamble to the poll. If the amendment passes, it will be understood as players rated above 2199, not above 2200.

No, No. When I read the amendment, I do not understand what the changes mean. It does not say that players over 2200 (or 2199) are not eligible for bonus points. I think that is Paul's point.

Anyway, if we all understand that YES means no bonus points for guys over 2199 (or 2200), when fine.

Michael von Keitz
04-05-2012, 11:03 AM
No, No. When I read the amendment, I do not understand what the changes mean. It does not say that players over 2200 (or 2199) are not eligible for bonus points. I think that is Paul's point.

Anyway, if we all understand that YES means no bonus points for guys over 2199 (or 2200), when fine.

Okay, I've pasted a note at the top of the preamble. I probably could have dealt with it differently, but I wanted the message as close to the poll as possible. I thought Paul's note was in reference to Lyle's "notice to governors" in this thread (i.e. Lyle's reference to "players rated over 2200"), so thanks for pointing out the misunderstanding.

Egidijus Zeromskis
04-05-2012, 11:05 AM
Anyway, if we all understand that YES means no bonus points for guys over 2199 (or 2200), when fine.

After re-reading I agree that the L.C. edition confuses my intention. The statement should be in the text ""Bonus points are not calculated for players rated above 2199"". (or "at or over 2200")

Lyle Craver
04-05-2012, 04:46 PM
It certainly was not my intention to confuse and I apologize to any one who was confused.

Paul Leblanc
04-05-2012, 07:59 PM
I think a "yes" vote means that my motion goes to a later vote with an amendment stating that 2200+ players would not receive bonus points.
This isn't a vote on my motion, it is a vote to amend the motion. If the amendment passes, we have to vote again on the amended motion. For example, I voted against the amendment but I may vote to implement the new, amended bonus point formula when we enter the voting phase of the meeting. Am I correct on this?