Scrabble New South Wales

National Ratings System

Adapted from an article in Across the Board, September 1993
The Australian Scrabble Ratings System was developed from the U.S. Scrabble Ratings System, which in turn was based upon the "Elo" chess ratings system developed in 1959. Here is how the system works:

Each player will have a rating usually ranging from 400-2200, with the higher number representing the better player.

The winner of a game gets a number of ratings points, which is the probability of the player losing, times their multiplier. Similarly, the loser loses a number of ratings points, this number being the probability of the player winning, times their multiplier.

This means:

Good players don't gain much beating lesser players
Lesser players don't lose much losing to good players
Good players lose a lot when they lose to lesser players
Lesser players gain a lot when they beat good players
When two players of unequal rating play, the probability of the higher rated player winning is assumed to be related to the difference in rating:

Difference of Ratings:050100150200250300400500
Probability of better player winning50%57%63%70%76%81%85%92%96%
Probability of worse player winning50%43%37%30%24%19%15%8%4%

From this table (and the ratings of your opponents) the system can calculate how many games you are expected to win in a given tournament.

If you win more than expected, your rating will go up, but if you win less, it will go down. (There are a few special cases which will be mentioned later)

Once the probability of winning has been calculated, the programme can work out how many rating points you would gain if you won.

Depending on how many games you have played, and your current rating, each player is given a 'multiplier'. The multiplier determines how "stable" a players rating is considred to be.

Multiplier Table
Rating1 to 10 games played11 to 49 games playedover 50 games played
2000+70 minus no. of games2010
1800-199970 minus no. of games2416
0-179970 minus no. of games30 20

The actual points gained or lost depend on the multiplier as well. The following figures are based upon the player being ranked under 1800, and having played more than 50 games. This gives a multiplier of 20.

Multiplier 20
Difference of ratings:050100150200250300400500
Points gained: Better player wins108.67.46.04.83.83.01.60.8
Points gained: Worse player wins1011.412.614.015.216.217.018.419.2

Bonus Points

The system also allows for 'bonus' points to be awarded to players who perform considerably better than expected in a tournament. These points are awarded if the player receives, for the tournament, more than 5 points per game. (ie 35 points for a 7 round event). The player who exceeds this is awarded 2/3 of the excess as a bonus.

Anyone who plays a player who receives bonus points will receive some 'feedback' points - since they played someone who was underrated. These feedback points are one point for every 20 bonus points, or part thereof, for each opponent.

In practice players receive around 50-75 bonus points when they perform very well. Most players receive between 0 and 10 feedback points in a tournament. There is no negative equivalent of bonus or feedback points. There is no loss of rating points for non-participation in tournaments.


Actual calculations are performed by computer. If a tournament is run using the 'Swiss' program, updated ratings are available within minutes after the completion of play.

The Australian National Rating System was conceived and designed by Barry Harridge, and agreed upon by CASPA at its inaugural meeting in 1992. Following this, Kevin Burfitt developed the computer program and circulated initial results.

The system has been operating smoothly since 1992, with some players reaching 1000 games in 1997. It is generally conceded, however, that the ratings are declining overall, and remedial programming will be necessary to arrest the fall, and perhaps increase overall ratings until they are on a level commensurate with North America and New Zealand.