World Othello Federation

[Lazard]

Hello.
Before going on to the subject, I would like to renew my apoligies to everyone for not having been able to work the past year on WOF as I intended to (and as people hoped). (For those interested in computing, the main reason is here : http://www.pearsoneducation.fr/espace/livre.asp?idEspace=75&idLivre=2715.)
I hope the new WOF will work hard to come up with new things and I would be happy to help anyone with my work and knowledge.

About ratings:
The current FFO ratinglist collects tournament results from all over the world (except Japan) and is used as an unofficial world ranking. I would like to propose some changes that I hope people will welcome.

As now, the ratinglist uses all results from the past 14 months, each game is given a weight of 1. Someone not having played in the past 14 months will drop out of the list and recent games do not count more than older ones.

I propose to use the following periods and weights :
Period 1 : 7 last months, weight 150
Period 2 : 7 months before, weight 100
Period 3 : 12 months before, weight 60
Period 4 : 12 months before, weight 30
Period 5 : all previous games since 1992, weight 1

This way:
- Recent games have more influence on the ranking,
- More games are included

But there is one drawback. As all games since 1992 are used, retired players appear in the list, often near the top. I don't think it's a a good thing for public relation (tell a journalist that Feinstein is the best English player... but that he doesn't play anymore...) so I remove from the list players that haven't played in the first 4 periods (38 months or just over 3 years).

You can compare current ratinglist and the new one at :
http://emmanuel.lazard.free.fr/Class.html

Please comment!

[Henry Aspenryd]

Congratulations to finishing your book Emmanuel!

When it comes to the rating, the weighting of the different timeperiods seems well proportion for the first 4 periods. What I do not understand is why games older than 3 years should be weighted so lightly? Have the game and challenge of Othello tournaments change so much during the years?
I have seen a couple of games played between top rated players from WOC's prior to 1990 and I must say that the quality of the games are quite high even though they had no programs to help them train. Games played 4 or 5 years ago certainly holds a high standard. Maybe a weight of 15 could be used for the old games?

To have an active period of around 3 years seems fair and reasonable. I guess that some would like to see a complete list without this constraint for nostalgic reasons. But that would be another list

I noticed that alot of players got a negative rating, presumably debutants that lost most or all of their games in one tournament. These players do not really complete a world ratinglist and I can't imagine that they feel very proud of having their names on the list for the coming 3 years. Could we in some way exclude these players from the list with any rule? If I remember correctly you have already excluded players that have no wins? Maybe we could say that players with uncertain rating (grey in the old and starmarked in the new rankinglist) that have a negative rating is excluded?

Lastly it might be easier to understand the periods if presented as:
Period 1 : 0 - 7 months, weight 150
Period 2 : 7 - 14 months, weight 100
Period 3 : 14 - 26 months, weight 60
Period 4 : 26 - 38 months, weight 30
Period 5 : all previous games since 1992, weight 1


Really good work with the ratinglist!

/Henry

[Lazard]

[Lazard]

The French Othello Federation has decide to use this new rating as its official rating.
It will soon be displayed on the FFO website at the usual address:
www.ffothello.org/classement/classement.php

[Ben Seeley]

The rating system appears to be better than before Good work!

I have an impression that although the older games are given less weight in the rating, they still have an equal weight in the error calculation? Not to potentially detract from the rating being more accurate, which is the most important thing .
_________________
www.othellogateway.com/blog

[Henry Aspenryd]

A thought crossed my mind. Before we introduced the different periods and weighting algorithm, all games where equal in weight. But now it is interesting not just to see number of games, but number of games in each period. Or maybe group them together in Period 1, Period 2-4 and Period 5, or something similar if there is problems with the space in the table.

[Roel]

As the FFO-rating list still does not cover the entire globe and can not yet be called a World Othello Rating, are there still plans to extend this rating system globally and make it more official?

[Henry Aspenryd]

In the WOF meeting in Athens as far as I know both Japan and US agreed to send all their results to the FFO.
If I am correctly informed it was even a demand on the federations that they should send national results in order to get 3+1 spots in the WOC.

[Tom_BOA]

[Tom_BOA]

and in the mean I'm in happy expectation since last Thursday to see 500 new Japanese games and more than 90 new players in the list

(everything I have 2 mails in my mailbox I hope its Manu with a notification )

[Lazard]

Thanks Tom.
I'm really busy at the moment so I haven't had time to work on the new files you sent me. But thanks for all your efforts.
I'll try to update the ratinglist next week.
Emmanuel

[Lazard]

The new rating is online!
Thanks to Tom who sent me 5 Japanese tournaments.

[Vladimir KHLEBNIKOV]

1. Why is used "e^x", but not "3^x".
When
1/(1+e^(-1))>3/4 but 1/(1+3^(-1))=3/4

2. The rating live its own life, outside of dependencies from player, when certain amount of the games goes from period at the following period. Because the weight of the games is discrete period by period: 150-100-60-30-1. This looks like drive on medieval road with hummock and pit. I offer drive on superhighway!
Weight of the game is 0.998^(T0 - Tg)
T0 -- a current date /days, not hours, minutes or seconds:)/
Tg – the date of the game.

Нарру New Year!
_________________
Othello foREVER!