[AusRace] MA Jockey Plan - a system

[Tony Moffat]

MA wrote ' Jockeys win occasionally, some more often than others' and ' a
jockey on a popular horse wins more than their share, more than they
should'. Popularity in this case is determined by the (pre-race) price it
owned.

MA wrote to a newspaper in Melbourne in the 60's explaining that winning on
the punt was as plain (s/he didn't say 'simple') as using the statistics to
'guide' you when selecting your bets each Saturday.

How about this, he announced, a jockey on an odds on runner can expect to
'salute' 40.7% of the time. The full table is as below, Jockeys long term,
can be expected to 'score' viz
Odds on - 40.7
Up to and incl. 5/1 - 20.57
Between 6/1 and 11/1 - 10.89
Between 12/1 and 25/1 - 5.43
Over 25/1 -  2.75

It is always seen (s/he didn't say 'often') that a jockey on an odds on
runner 'salutes' more often than his recent win/loss record shows he might -
the horse helped, and so it goes with all runners

It is likely that the possibility of the jockey performance at odds, and the
natural jockey win/loss record could be combined to an extent where the
probability of success is enhanced.

MA suggested the jockeys win percentage and the horse score, decided off the
table, be averaged and a betting choice made from that - big numbers help

I looked at this method of selection. It did select winners, and place
getters but perhaps no more  than that using the market as a guide
ordinarily.

I modified the figures by normalizing the odds table (to 100%- *1.24) and
using the (b) fraction from that to multiply the jockey PLACE percentage and
while the result stayed around the same selections, some others, wider in
the market were now included, which is a benefit.

Lastly I used the horse place percentage, the jockey place percentage, and
the information from the table in an equation to get a selection - it works
also. What is evident is the lack of success of horses out beyond $26 -
lonely.

There is an apparent doubling (almost) in the values in the table 2.75+2.75
= 5.5 (5.43), 5.43+5.43 = 10.86(10.89) - I don't know where/how arrived at
his data but it looks good especially when it seems Malcolm Knowles and
others did a big examination in the early 90's and they too arrived at these
figures also.

Is there a database guru who could prove if the numbers hold up?

Cheers

Tony


-- 
This email has been checked for viruses by AVG.
https://www.avg.com