Distance preferences – more to do with mentality than physiology

While analysing the performance of the Race Modlr computer model in different types of races, something that became apparent was the fact that the model’s analysis of horse’s distance preferences was adding virtually nothing to the accuracy of its race predictions. This raised the question, why?

That doyen of thoroughbred breeding Federico Tessio staunchly believed that good horses were fast horses and that a fast horse would beat a slow one whatever the distance. Our mathematical model of how horses fatigue in relation to their ability, based on treadmill research into their physiology, largely supports this. It suggests that Frankel was physiologically capable of winning a July Cup over 6f or one of the Cup races over 2m. Race Modlr’s computerised handicap rated Frankel a 145 over 1m and throwing that into the aforementioned mathematical model suggests he’d have been a high 120’s sprinter or stayer.

It infers that distance is much less of a factor in determining the outcome of races than pundits frequently make out. The reason for this is likely to be that the range of distances over which horses race is actually very compact when compared with humans. In human terms a sprint is 100m to 200m, a middle-distance race 800m to 1500m and a staying race 5000m up to 26 miles. Physiologically these require very different body compositions, sprinting making use almost entirely of fast twitch muscle and anaerobic respiration (energy production without oxygen), staying using lots of slow twitch muscle and being almost entirely aerobic (energy production requiring oxygen).

If we compare how slow humans go at the shorter end of middle-distance races i.e. 800m with how fast they go in sprints and compare that with how fast a horse can run and how fast it actually runs over the various race distances horses contest, we quickly see that physiologically the extremes of horse race distance are not that different. We know a human flat-out can manage 100m in 9.5s. The average speed for the 400m world record is 16% slower than at 100m and at 800m 32% slower. Plotting this graphically, we get a curve looking like this:

http://www.theracingapp.co.uk/blog/wp-content/uploads/2015/09/Screen-Shot-2015-09-29-at-13.43.58.png" rel="nofollow">

We also know that a racehorse flat-out could manage a furlong in about 10s. The average speed over 5f is 15% slower than this, but at 8f it is only 22% slower, at 12f only 25% slower and at 2m only 29% slower. If we use the equation that defines the curve seen on the chart to work out what that means those race distances equate to in human terms we find the following:

5f = 300m

8f = 415m

12f = 476m

16f = 560m

Even the 3m 2f Cheltenham Gold Cup is only roughly equivalent to a human 1500m race, albeit a steeple chase.

Excluding Usain Bolt, who is something of a freak, were there a 300m world championship for humans then the 200m champion would probably be favourite. Over any of the other distances mentioned, 415m, 476m or 560m then the world 400m champion would probably be favourite. What this suggests is that in reality only two horse types are required on the flat, sprinters and non-sprinters because apart from 5f and 6f the physiological requirement of all the other distances is much the same. The ability of a 7f horse to stay 1m 4f is more about its mental make-up than its physiological make-up, since provided it settles then physiologically it will stay.

Interestingly, what this confirms is that by accident or design the evolution of the the English Triple Crown which tested a horse’s speed over 1m and its ability to settle over 1m6f was in all likelihood a truer test of the racehorse than the current focus on racing between 1m and 1m2f. Today’s approach simply repetitively testing the same thing.

Our approach to predicting distance preferences in Race Modlr has revolved around retro-engineering the work conducted on the equinome project where horses have been identified with three genetic variations for distance preference. The equinome project confirmed the presence of these genotypes by looking at the best distance of horses exhibiting them. Conversely, since we know the best distance of each horse from our computerised ratings we can predict the probability of each sire and dam being of a given genotype, and consequently the probability of their progeny being of a given genotype. We can then predict the probability of any given distance being a horse’s best.

Testing the accuracy of these predictions against all horses aged three-years-old and upwards that have raced on the flat in 2015, we found a strong relationship between the average observed best distance of a group of horses and their predicted best distance. Unfortunately, accurately predicting individual horses to within one furlong seems almost impossible with a success rate of about one in six. What holds much more statistical validity is separating the sprinters from the non-sprinters. Of horses with a predicted best distance of less than 1m, 70% produced their best form over 7f or less. The one horse prominent in the betting for next year’s 2,000 Guineas this pertains to is of course Shalaa.

At the other end of this research is Gleneagles. Our model actually considers him to have more potential to stay than Jack Hobbs or the 2-y-o Deauville, second favourite for next year’s Derby. Given that Gleneagles always seem to settle beautifully, you have to wonder what might have been over further.

Golden Horn, who today added victory in the G1 Prix de l’Arc de Triomphe to his triumphs in this year’s Derby, Eclipse and Irish Champion Stakes, and the highest-rated turf horse in the world, will become a Darley stallion in 2016 and will stand at Dalham Hall Stud. A fee has yet to be announced.

Anthony Oppenheimer’s homebred son of Cape Cross made the perfect start to his racing career, setting a juvenile course record at Nottingham when breaking his maiden on his only start at two.

Golden Horn put up another record-breaking performance first-time-out at three in the Listed Feilden Stakes at Newmarket, knocking some three seconds off the previous Stakes record. He then stormed into the Derby reckoning after an impressive win over subsequent Irish Derby victor Jack Hobbs in the G2 Dante Stakes at York, drawing almost three lengths clear.

His victory in the highest-profile Derby trial prompted Mr Oppenheimer to supplement him for the Epsom Classic, and his faith was rewarded in scintillating fashion as Golden Horn powered home to take the race by three-and-a-half lengths in the third-fastest time ever, Jack Hobbs again the runner-up. The victory earned Golden Horn a Timeform rating of 130, fully 4lbs ahead of Cape Cross’s first Derby winner Sea The Stars after his own brilliant win in the race.

Dropped back to ten furlongs for his next start in the G1 Eclipse, Golden Horn showed that he could also master his older rivals as he made all and saw off a strong challenge from multiple G1 winner The Grey Gatsby, once again winning by three-and-a-half lengths.

After a narrow second on rain-softened ground in the G1 Juddmonte International at York, he bounced back to his brilliant best in the G1 Irish Champion Stakes at Leopardstown last month.

Golden Horn’s sire Cape Cross has a long connection with Epsom’s greatest races: besides his two Derby-winning sons, he has also sired Oaks winner Ouija Board, herself the dam of Derby winner Australia. Out of Fleche d’Or, a half-sister to G1 Coronation Stakes heroine Rebecca Sharp, Golden Horn is from a family that has been nurtured by the Oppenheimer family at their Hascombe And Valiant Stud for half a century.

Sam Bullard, Director of Stallions, said: “It is wonderful that Darley has bought into this great Champion and we are thrilled that Mr Oppenheimer has decided that Dalham Hall Stud is where Golden Horn will get the best opportunity of becoming the great stallion we believe he can be.”

by Makfi - Zingo, by Zabeel