- Byron Rogers
What are the Odds - Odds Ratio's at Yearling Sales
We are in Australia at the moment for the yearling sales series which kicks off next week at the Magic Millions in Queensland. In fact, the Magic Millions Sales company bookends the Australasian yearling sales series with the National Yearling Sales held in late May generally considered the last of the yearling sales of the year.
We have a new announcement to make later this week on a new product developed for the yearling and two-year-old sales which we are really excited about, but prior to that we've been busy getting clients organized to make sure that we can get orders filled. There are a number of different requirements, depending on the client, but one of the more common themes of the conversations we've had with them seems to be "where is the best value to be found?"
It's a good question. After all, the sales series doesn't fall in any order of merit and there is an underlying belief that a 'good horse' can come from any sale. The first sale, the Magic Millions at the Gold Coast, is presumed to be one of the more profitable sales to buy at in terms of producing graded/group stakes winners while the very next sale, the Inglis Classic Sale is presumed to be one that is good "in patches". Inglis Easter is traditionally where the best pedigreed horses are found, but does that mean we should be holding back cash for that sale over others?
Obviously "value" is somewhat subjective, but for us it is defined by the group winners that come out of a sale (Listed winners aren't generally that much different to a multiple allowance/city winner) and we decided to take a look at it and see where the "value" is to be found. Honestly this is probably a two tiered answer with the first to establish where the best horses are found, and then secondly working out in which price range the value is found.
There are a couple of ways to answer the first question, but one of the best ways to do this is to look at Odds Ratio's. Odds ratios (OR) are used to compare the relative odds of the occurrence of the outcome of interest (e.g. becoming a group stakes winner from a sale), given exposure to the variable of interest (e.g. being offered at a sale). The results can be interpreted as:
OR=1 Entry into the sale does not affect odds of becoming a GSW
OR>1 Entry into the sale is associated with higher odds of becoming a GSW
OR<1 Entry into the sale is associated with lower odds of becoming a GSW
Odds ratio = (A/C) / (B/D)
A= Positive Outcome in 1st Group B= Positive Outcome in 2nd Group C= Negative Outcome in 1st Group D= Negative Outcome in 2nd Group
So for our purposes of comparing one sale against all other sales,
A= Number of GSW in the subject sale B= Number of GSW in all other sales C= All Offered in the subject sale D= All Offered in all other sales
We gathered the sales results outcomes from the 2008-2011 yearling sales for the major Australasian Sales. Here are the numbers:
So what does this table show?
Firstly, and unsurprisingly, you are far more likely to find a Group Stakes winner at the Inglis Easter Yearling sale than any other sale - significantly so. NZB Karaka and the upcoming Magic Millions sale also provide some value in terms of group stakes winners being offered at those sales over all other sales. Inglis Melbourne is just about a wash, while the rest of the sales put the buyer at some disadvantage when compared to other sales in finding a group stakes winners.
Buyers make the a priori assumption that the yearling sales are in some way selected by the sales companies and consignors themselves and the data above seems to suggest this. Of course there is also a high trainer effect in these figures as the better trainers tend to go to these three sales and buy their horses there so it may in some ways be a self fulfilling prophecy. What the data also does tell us is that if we have a client wanting to buy four yearlings, they would be best buying one at Magic Millions, one at Karaka and two at Easter - where the odds are in your favour that a Group Stakes winner can be found.