|
Yorkshire
Chess Association |
Last update: |
|
|
The 40
Point Grading Horizon |
|
The points scored by a player
in the grading of a chess game are dependent on the opponent's grade, but only
within limits.
The difference between
people's grades represents the approximate difference in their expected
percentage performance against players of each other's grade (see The Meaning of Grades). However, simply adding or subtracting 50
grading points for a win or loss is an arithmetical simplification which made
life relatively nice and easy when grading was done manually, before computers
started being used.
The
Problems Causes by Simplification
In the language of
statistics this percentage interpretation of grades approximates a sigmoid to a
straight line, or a standard distribution curve to two straight lines. In layman's terms it assumes you will always
beat a player 50 points weaker than you, which is untrue, and means you will
score more than 100% against even weaker players, which is
meaningless. The resultant simplified
calculation would mean you'd score only your own grade for beating an opponent
graded 50 points lower than yourself, which is not too unreasonable, but would
score less than your own grade for beating to a player more than 50 points
lower, which would erroneously imply you were playing below your grade.
The Simplest Solution
It would make more sense to
treat the opponent as being 50 points different from ones own. That way you wouldn't lose grading points
for beating much weaker players, nor would you gain grading points for losing
to much stronger opponents. Such a
50-point horizon would still cause problems.
Imagine a group of players all of around 110-grade strength who only got
the opportunity to play each other. If
none improved or got worse then they would all tend to get new grades still
around 110. That’s perfectly okay, but
what if one player were to improve significantly from year to year, winning all
his games? With opponents all graded
around 110, a 50-point horizon would not allow him to rise above 160, even if
he were reaching grandmaster strength!
A Better Solution
By making the grading
"horizon" less than 50 points (e.g. 45, 40 or 35) such improving
players would not be tied down to some ceiling above the grading level of their
opponents. The other side of the coin
is that the grade of the player in the example just given would in time rise
beyond his true strength, but this is an artificially extreme example.
The grading system is
considered to work best with a horizon less than 50. The limit fixed was 40.
This is a somewhat arbitrary but pragmatic choice. 37 or 42 might be "better" choices
but would have made mental arithmetic that bit more taxing and error prone.
What This Means in Simple
Terms
Thus if you have a grade of
say 118, then your games will be graded as follows.
|
opponent's |
result |
points |
calculation |
|
165 |
win |
208 |
opponent |
|
(over 40 |
draw |
158 |
treated as 40 |
|
higher) |
loss |
108 |
higher then self |
|
158 |
win |
208 |
|
|
(40 points |
draw |
158 |
opponent's |
|
higher) |
loss |
108 |
grade plus 50 |
|
130 |
win |
180 |
for a win |
|
(less than |
draw |
130 |
|
|
40 higher) |
loss |
80 |
|
|
118 |
win |
168 |
opponent's |
|
(same |
draw |
118 |
grade |
|
grade) |
loss |
68 |
for a draw |
|
100 |
win |
150 |
|
|
(less than |
draw |
100 |
|
|
40 lower) |
loss |
50 |
opponent's |
|
78 |
win |
128 |
grade less 50 |
|
(40 points |
draw |
78 |
for a loss |
|
lower) |
loss |
28 |
|
|
70 |
win |
128 |
opponent |
|
(over 40 |
draw |
78 |
treated as 40 |
|
lower) |
loss |
28 |
lower then self |