PDA

View Full Version : A Ratings System



jase
11-04-2004, 09:26 PM
One of the schools I work at is a large-ish high school with a healthy chess playing base [about 40 students attend chess class every week, and at least that many again play socially during the week].

I am thinking about implementing a rating system within the school. There's several reasons for this, which I can go into if anyone's interested. Briefly:
- few have NSW Junior Chess League ratings
- create an extra incentive
- offer a measuring stick within the school

My question is, how to do this? Is there a simple program that I can download? I have SwissPerfect, which has some ratings calculations options, but it doesn't appear to extend beyond offering a performance rating for each player in a tournament.

I was going to ask this in a PM to Bill Gletsos, but I thought there were others [including Bill of course] that might have some resources or suggestions.

Muchas Gracias,
Jase

Alan Shore
12-04-2004, 05:53 AM
One of the schools I work at is a large-ish high school with a healthy chess playing base [about 40 students attend chess class every week, and at least that many again play socially during the week].

I am thinking about implementing a rating system within the school. There's several reasons for this, which I can go into if anyone's interested. Briefly:
- few have NSW Junior Chess League ratings
- create an extra incentive
- offer a measuring stick within the school

My question is, how to do this? Is there a simple program that I can download? I have SwissPerfect, which has some ratings calculations options, but it doesn't appear to extend beyond offering a performance rating for each player in a tournament.

I was going to ask this in a PM to Bill Gletsos, but I thought there were others [including Bill of course] that might have some resources or suggestions.

Muchas Gracias,
Jase

If it's just within the school, I would suggest an ordinal ranking system, or 'ladder' may be the best idea for you - a lot simpler and it's been implemented before in other activities such as tennis with success.

jase
12-04-2004, 12:32 PM
Thanks Bruce. I have used Ladder Systems before, and they are okay. However with tennis [and golf] it's a kind of ratings system anyway, with allocated points per tournament producing a "ladder".

I really want to implement a rating system. I think it will be fairer, more accurate, and create more of a sense of bettering yourself. I have found a couple of sites with elo calculators, but nothing yet that enables me to maintain a ratings database. I did find www.schoolchess.com which seems to have a programs to manage ratings, but it'll cost me $US99, so bugger that.

If possible I even want to create my own parameters. For example a top rating of 1000, with perhaps 4 or 5 published class levels down to about 200.

Bill Gletsos
12-04-2004, 08:47 PM
jase,
Firstly it is virtually impossible to rate a group of players if none of the players already have a rating.
Even if you have a few rated players you need to get the unrateds to play the rateds a sufficient number of times.
If the unrateds play a few rateds but many unrateds without iteration of the results it will take ages for the unrateds to get a semi accurate rating.

Therefore it may be wise to run some sort of scheveningen tournament where your rateds play a group of unrateds. You will need to repeat this a number of times.

BTW I have not seen any publicly available rating software.

jase
12-04-2004, 10:05 PM
jase,
Firstly it is virtually impossible to rate a group of players if none of the players already have a rating.
Even if you have a few rated players you need to get the unrateds to play the rateds a sufficient number of times.
If the unrateds play a few rateds but many unrateds without iteration of the results it will take ages for the unrateds to get a semi accurate rating.

Therefore it may be wise to run some sort of scheveningen tournament where your rateds play a group of unrateds. You will need to repeat this a number of times.

BTW I have not seen any publicly available rating software.


Thanks for the reply. I have a simple solution to the lack of ratings for inter-school rating purposes: I am going to give them all a provisional rating. A majority have played 3 or 4 interschool tournaments over the past 2 years, so I have some detail when assessing their playing strength.

Are you at liberty to inform what software our ratings officers [both junior and adult] are using?

Bill Gletsos
12-04-2004, 10:12 PM
Thanks for the reply. I have a simple solution to the lack of ratings for inter-school rating purposes: I am going to give them all a provisional rating. A majority have played 3 or 4 interschool tournaments over the past 2 years, so I have some detail when assessing their playing strength.

Are you at liberty to inform what software our ratings officers [both junior and adult] are using?
I dont know what software the junior rating officers use. You could talk to Alan Wright(NSWJCL rating officer) or David McKinnon who runs the QLD junior list).
The state rating officers dont run any rating software. They just submit Swiss perfect files. All they need to do is ensure that players are on the ACF master file and that id numbers are in the SP files.

The ACF rating program is too complicated for your task.

PHAT
05-03-2006, 08:41 PM
Firstly it is virtually impossible to rate a group of players if none of the players already have a rating.


This is totally wrong and you are a fool. You have no idea what you are doing. You are a liability and nobody should ever ask you for help with anything other than to give you advice on where you should go.:hand:

Bill Gletsos
05-03-2006, 09:58 PM
This is totally wrong and you are a fool. You have no idea what you are doing. You are a liability and nobody should ever ask you for help with anything other than to give you advice on where you should go.:hand:As usual you totally have no clue.

jenni
30-04-2006, 10:41 PM
Thanks for the reply. I have a simple solution to the lack of ratings for inter-school rating purposes: I am going to give them all a provisional rating. A majority have played 3 or 4 interschool tournaments over the past 2 years, so I have some detail when assessing their playing strength.

Are you at liberty to inform what software our ratings officers [both junior and adult] are using?

Alan Wright gave Canberra a copy of his ratings software some years ago and we used it to create a junior rating system in Canberra. From memory it was pretty easy to run - I think Alan had developed it himself.

When the ACF rapid rating came in, we abandoned it and used the ACF system instead. Had you thought of doing that?

Southpaw Jim
30-04-2006, 10:41 PM
Are you at liberty to inform what software our ratings officers [both junior and adult] are using?

This is probably a bit useless given what Bill said about rating in the absence of a number of rated players - could you implement the Glicko formulae in MS Excel? (or another spreadsheet of choice)

For the reasons given by Bill, the rating differentials would probably take ages to come down to a reasonable level, but no harm in trying I guess.

Garvinator
01-05-2006, 06:07 AM
When the ACF rapid rating came in, we abandoned it and used the ACF system instead. Had you thought of doing that?
especially now that the rapid system is free;)

jenni
01-05-2006, 11:07 AM
especially now that the rapid system is free;)

Well that is what I was thinking. :)

Kevin Bonham
05-05-2006, 02:10 PM
You can use an assumed mean rating and iteration to get a very rough figure for a bunch of unrated players from a large Swiss or round robin where they have all played lots of games and each has played a fair proportion of the others, and use this as a start point for a system for those players.

By iteration, I mean start with all players having the same rating and use the tournament results to get a first performance rating for each on that basis. Then use the first performance rating as the rating for each player and use the tournament results to get a second performance rating for each on that basis. I'm not sure what the optimum number of iterations to stop after is - too few and the spread of the ratings is too conservative, too many and you get some very radical differences that reflect the small sample size of games. I would stop after two or three iterations to seed the system.

This sort of thing is only recommended for very rough informal level situations where people's interest in having a number after their name exceeds their concern about their rating being easily 200 points out. I experimented with using a 14-round Swiss event with 30 unrated players to see how such a system would work once. It was reasonably satisfactory although the problem with really weak unrated junior events is that you get players who are weak enough that they really deserve a negative rating.

four four two
05-05-2006, 02:16 PM
By negative rating I presume you mean a fairly low one,less than 400.:hmm:

Kevin Bonham
05-05-2006, 02:49 PM
No, I mean one actually less than zero. In the past there have actually been players with negative ACF ratings. There are currently some with ACF ratings below 200 but there are unrated players in schools out there who would not score 20% against unrated players who would not score 20% against the under-200s.