Unexposed Pedigrees & the Explainability Paradox

Unexposure in racing is something we talk frequently about in terms of potential improvement in horses. As the form of horses becomes more exposed, the less they win races overall. Especially in handicap driven jurisdictions like the UK & Ireland.

Horses having their 3rd career start running in a handicap win at a 10.15% return, compared to 7.74% by the time they have their 50th start. The graph below shows the win strike rate of horses in UK & Ireland Flat Handicaps by Career Run.

It’s even a decent betting angle, as those unexposed 3–10 career start horses in handicaps have won +189.78 races in UK & Ireland handicaps since 2008 than would be expected by their market starting price. The market expected them to win 32,246 races and they returned 32,436.

That type of angle led me to thinking about we could qualify “unexposure” within a pedigree. And if there could be a similar edge vs. the bloodstock market by employing such a concept. To evaluate we utilised a dataset of 269,239 yearlings sold between 2012 and 2022 across the UK, Ireland, USA, Australia, France, Germany, Japan, New Zealand.

This is global data, but i’ve adjusted all prices back to GBP for comparative analysis purposes.

Purchasing Preferences

Let’s start by evaluating whether the market has any specific purchasing preferences based on foal number. We found slight differences here in Sale Status (Passed in / RNA vs. Sold) between the foal number; 76.52% of first foals were sold with a drop to 72.29% of 10th foals. I had actually expected the market to take more of a suspicious view on first foals and Pass In a higher proportion than the later foals, interesting to see that theory debunked.

First foals returned the lowest average purchase price in our dataset at £47,975 per foal, peaking at £58,532 by the 9th foal. There is an inherent selection bias there, as the mare still commercially selling by the 9th foal has likely been a good and proven producer.

Racing Performance

Evaluating our foal counts then in terms of subsequent racetrack performance outcomes. We observed that first foals had a higher % return of making the racetrack; 76.87% of the first foals in our dataset made the track vs. 61.90% of 10th foals. This relationship continues through all layers of performance.

First Foals returned as Winners at 50.49%, we observed a marked drop to 42.90% by the 6th foal, dropping to 33.30% by the 10th foal.

The first 3 foals out of the mare (Foal Number 1–3) returned 5984 stakes winners from 197,591 yearlings offered. That’s a return of 3.02%. Those mares still producing by the 8th — 10th foal returned 99 stakes winners from 5878 yearlings offered, a return of 1.68%.

The Explainability Paradox

Now we come to the explainability paradox. The buyer of bloodstock will often have a client to whom they will have to justify the purchase of said stock; be that a Bloodstock Agent explaining to their client, or a Trainer or Syndicator evidencing to their prospective ownership base. It is easier to explain a purchase to a prospective client based on the dam being an already proven producer with winners on the page. Our hypothesis would be that by that point you run the risk of already getting towards diminishing returns, especially beyond the 5th foal.

This isn’t to say i don’t think you should pay attention to the record of the mare as a producer — in fact, it’s actually one of the top metrics in performance modeling in bloodstock. But more to say that “unexposure” in itself is also a factor, and a particular value factor at that. (I use %RivalsBeaten across all runs by Mare Progeny as an important variable in pedigree modelling — a mare with <5 foals who is a proven high quality producer already is a big advantage)

Let’s look at how much you would have had to turnover per winner, per stakes winner and per group winner as a way of evidencing market value based on foal number.

On the first foal, the buyer had to turnover £72,705 for every winner returned. That’s the total sales spend in GBP divided by the total number of winners in the dataset. By the 10th foal the buyer is turning over £111,800 per winner. The performance outcomes have declined, but the average price per yearling has increased — the market moves in the wrong direction.

Buyers had to turnover £1.29m for a stakes winner from the first foal population, comparative to £3.31m for a stakes winner in the 10th foal population. The first 3 foals return a stakes winner every £1.23m worth of spend vs. £2.6m for foals 8, 9 and 10. You have to outlay double the investment to return a stakes winner from the later foal populations than the earlier ones.

The premise of this article was to explore the concept of “unexposure” within a pedigree and whether it had any relationship to racing performance or market value. What we can see from the table above is that on a population level, the more “exposed” a page is on the dam side the lower the performance returns and the higher financial outlay required.

The contrarian buyer who is willing to look into purchases out of unexposed mares may be able to find a value edge within a market driven by explainability.