Thanks to Howard Grubb for the following:
The points appear to be derived from regression models, presumably fitted
to data, though I don't know what the performances are that were modelled.
The general formulae of the models are:
points=a(b-result)^c for running (better results are less time)
points=a(result-b)^c for jump/throw events (better results are more dist)
e.g. a=25.437, b=18, c=1.81 for 100m
a=0.1435, b=220, c=1.4 for LJ
see e.g. http://www.cs.uml.edu/~phoffman/mod8.html
or http://www.wava.org/wava appendix b.htm (with spaces)
(Howard's own webpages on athletic data
http://www.reading.ac.uk/~snsgrubb/athletics/ are well worth looking
at too).
However, a good answer leads naturally to further questions:
What was done before 1985?
How did they decide on the formulae? (It is a nice example of a
regression application if that's true.)
Who decides on when and how the figures are updated? (Howard thinks,
for instance, that javelin and discus were updated two years ago).
I suspect that there should be IAAF documentation on this but being
totally unfit myself am not the best person to know where to look!
Thanks for any help
Antony Unwin
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|