roadrunner wrote:Thinking about it further, surely this is a good test to put the algorithm through. Plainly, anyone who had an hour or more added to their time should be an outlier, so a good algorithm would discount them.
No
The algorithm doesn’t “know” what a good time for a sprint race is, or even that this was a sprint race – it’s simply given two lists of numbers: previous ranking scores and time in seconds, for each of which it calculates means and standard deviations, and then uses these to allocate points.
Also an outlier is defined as someone who’s score is 100 points greater or less than their average for the year, not according to their time on the course.