Page 52:
"We now wish to utilise the theory we have previously discussed in the context of the
problem of producing abnormal returns in a horse race betting market. Our application
will be purely focussed on the first task of finding estimates for the probabilities of victory
for given horses in a particular race. The data we use is somewhat contrived: it consists
of one hundred races where in each race there is exactly ten horses (not necessarily the
same horses appear in every race, however) 9. For each horse j in race i there are certain
characteristics associated with them. These include:
Their closing odds (i.e. the odds set at the start of the race) (a)
The days since their last first placing in a race (b)
The logarithm of their track probability, adjusted for the bookmaker's margin (c)
Some unknown variable (this particular variable's origin was not made clear to the
reader! (d)"
Last edited by traynor; 03-30-2017 at 10:04 PM.
|