retriever
Stallion
I see what you saying there....a 1st in one type of race would obviously be different to a 1st in another race type.For instance, A horse winning a selling race would get 5 points and horse placed 6th in a group race beaten 4 lengths would get 0 points. In the unlikely event that the two horses met in a race who would come out on top out of the two.
If one was to attempt 'improve' the ratings I think, for starters, we would have to widen the range. Say we used the full version (0-20pts) and widen to 0-100.
The max 5pts for LR would be 0-25.....2LR 26-50......C would be 51-55.....D would be 56-65.....rating and forecast the remainder.
We could add in other attributes within this range (or move ranges) as we see fit, or gauge how strong each attribute should be.
So, for example, if finishing position is deemed more important than CD - finishing position could operate from 0-40 and CD from say 41 to 50 and so on.....
Then we would have to gauge the strength of the LR win, LR 2nd, LR 3rd....etc. (and also, 2LR figures too) based upon the type race previously contested.
Perhaps just using 5 different 'race' types for placings, 5 different previous Course strengths, 5 different Distance strengths.......which would mean placing racecourses into set groups or grades.
I expect the code can easily be written to assess all this (IF, THEN, ELSE), so it's a matter of working out what's to be included, with what allocation of points and the perceived strengths of all the attributes.
Last edited:
As you rightly say the real testing is how these vendors use there own products .........assuming they do. I have been a Raceform user for 45yrs and its interesting to note that on several occasions during this period they have started specific tipping lines all of which failed.
