Generation intervals in the French trotter

It is generally accepted that faster generation intervals produce better trotters. This is particularly true for American horses. Quite a few breeders in Europe are skeptical of this and it is my experience that especially French breeders are dismissive of generation intervals altogether. But what is fact, and what is fiction? If we accept that generation intervals matter then surely it should matter for French trotters (Trotteur Francais) too, right? If faster generation intervals is better, then I expect horses with faster generation intervals to earn more and I also expect the starting percentage of these horses to be higher than for horses with slower generation intervals.

To avoid confusion: the generation interval indicates the time, usually expressed in years, between each generation in the pedigree. When there is less time from one generation to the next, i.e. when a horse's parents are quite younger, the generation interval is said to be faster. To only look at one generation when calculating the generation interval is not accurate as it ignores the overall generation intervals in the whole pedigree. I will use the method I think is best, the Sophia-50 number as generated by the Sophia pedigree program. The Sophia-50 number is found by counting all ancestors in positions in the pedigree born within 50 years of the horse's year of birth (so if the horse is born 2010, then all horses in the pedigree born 1960 and later are counted). If a horse does not have a complete pedigree going 50 years back from his birth, then he is excluded from the statistics.

The more horses analyzed, the more accurate the results are. So I analyzed all French trotters (Trotteur Francais) from 1990 to 2005. A total of 186 309 TF horses were born from 1990 to 2005. Since the Franco-American protocol was in effect to 1992 this excluded a few such horses (e.g. Defi d'Aunou etc) who are registered as US born but qualified for "dual studbook citizenship" under the protocol but there are fewer than 50 such horses. A total of 183 487 horses have a valid Sophia-50 number and are included in the analysis

Quick analysis of the 16 crops produced the following statistics:
Horses analyzed 186 309
Starters       58 859
Start % 31,59 %
Total earnings  2,667,941,841 €
Earnings/horse  14,320 €
Earnings/starter 45,328 €
€500k horses 331 (0.00 %)
€100k horses 7093 (3.81 %)
€50k horses 14936 (8.02 %)

(For the record, French born horses that are exported are also included - these figures are not limited to earnings in France)

If we start to look at the generation interval statistics
Lowest Sophia-50 5
Highest Sophia-50 85
Average Sophia-50 31,88

We see that the average generation interval is indeed quite slow with the highest Sophia-50 number being 85. In comparison, some foals of Explosive Matter have Sophia-50 numbers above 200.

To list every Sophia-50 number and the corresponding earnings can make you lose track of the bigger picture so I have summarized the numbers in the following table
Sophia-50 numbers Horses Starting percentage Average earnings
10 and below 41 24.39 % 3,158 €
11-20 14957 28.52 % 12,007 €
21-30           72747 29.83 % 12,706 €
31-40          66097 32.04 % 14,638 €
41-50         23853 36.05 % 17,763 €
51-60           4910 41.87 % 24,538 €
61-70          771 43.97 % 24,737 €
71-80 101 52.48 % 36,698 €
81 and above   10 60,00 % 17,619 €

Remember the averages calculated above. The average Sophia-50 number was 31,88, the average starting percentage was 31,59 % and the average earnings was 14 320 €.

All Sophia-50 numbers (or groups of numbers) below 30 have starting percentages and average earnings below the average for all trotters.
All Sophia-50 numbers above 40 have a starting percentage above average and an average earnings above the average for all trotters.

Looking at individual Sophia-50 numbers, for all Sophia-50 less than 32, both the average earnings and start percentage is below the average for all French trotters. For all Sophia-50 above 32, the starting percentage never falls below the group average. A starting percentage above 40 % is only found when the Sophia-50 is above 50. These numbers are averages, so a single horse with a Sophia-50 number of 20, meaning a very slow generation interval, can still be a good racehorse but it is an exception. On average, a higher Sophia-50 means a higher probability of the horse racing and a higher expected income. Similarly, average earnings above €20 000 is only found in horses with Sophia-50 above 49. The average earning for horses with Sophia-50 of 50 or more is 24,293 €. Only 6930 horses, 3.78 % of the TF born between 1990 and 2005, has a Sophia-50 number this high. On average, this group is doing much better than the rest - the average earnings are in fact almost double the average.

The trend is very clear: higher Sophia-50 numbers, meaning faster generation intervals, has a clear and marked positive correlation with earnings in the French trotter. Horses with slower generation intervalls can be better than a horse with a comparatively fast generation interval but on average the latter can be expected to be better if we look only at this factor. In reality other factors matter and breeding on quick generation intervals should not be done instead of breeding on quality individuals. But the statistics clearly indicate that French breeders should start paying attention to the generation intervals: when expected average earnings almost double as the Sophia-50 number goes from 30 to 60, that should translate into more expected money to the breeders.

A common objection to this conclusion is that the best mares usually get new and exciting stallions. I am doubtful that this objection is true. Many young mares are sent to established stallions with a proven stallion record. Moreover, some young stallions fail and the older stallions that are used are usually the ones who have produced. It does not rhyme with the conclusion to have young stallions failing and older experienced stallions succeeding. The statistics should be different, then.

What are the practical implications? It is easy to interpret this as me saying you should always pick younger stallions and younger broodmares. To repeat myself: that is not what I am saying. But, if you have a young broodmare and you are considering two stallions that you otherwise rate as equal you would be well advised to pick the one that gives the foal the fastest generation interval. Furthermore, if you have an older broodmare it makes a lot of sense to pick a younger stallion. Given the clear trend found above, it makes very little sense to breed old stallions to old broodmares.

In fact, if we think about it a little bit these findings provide a credible theory why many older stallions seem to produce worse as they get older: if these stallions do not get younger broodmares then each new crop will have progressively slower generation intervals, thus reducing both expected starting percentage and expected earnings. It is not the stallion that has lost his quality - instead the stallion needs a slightly different set of mares! I bet few, if any, stallion managers pay any attention to this but the statistics clearly indicate this may be a huge mistake.

Finally, even though it is tempting to draw the conclusion that "the faster the generation interval (=higher the Sophia-50 number) the better. Based on other analyses I would rather conclude that that "the slower the generation interval the worse it is". It is more crucial to avoid the really slow generation intervals than it is to make an already fast generation interval even faster.

Marloes Harkema is the manager of Sophia Pedigrees and a pedigree researcher/analyst. She can be contacted at