Breeding works with coincidence

When two individuals of different genotypes are crossed, it is not that important what the F1 looks like, as it represents maximum heterozygosity that is possible between two breeds. On each locus, it will have one allele from breed A and one from breed B while the goal of targeted breeding is the opposite, it wants to unite the desired traits of A and B on each locus homozygous, therefore, to stabilize the desired characteristics of the parental breeds in the new line. Therefore, the looks of an F1 are not actually all that relevant. In second-generation crosses, it becomes more complicated. Genes (or actually: chromosomes) get passed on entirely by coincidence, therefore there are numerous possibilities what a true F2 will look like, with a pure individual of breed A or B being the ends of the extremes. At this very early stage of crossbreeding, it would be a quite big coincidence if two F2 would happen to look exactly the same. I have an example from the Lippeaue here: 

 

Lale (Heck x Sayaguesa) x (Heck x Chianina)

Latina (Heck x Sayaguesa) x (Heck x Chianina) © Matthias Scharf

Both cows are (Sayaguesa x Heck) x (Chianina x Heck), yet they look very different. Lale has comparably long horn, a Heck cattle-like appearance overall except for the head, plus a slightly diluted coat colour. Latina, on the other hand, has the long-legged stature of Chianina and also the head and very small horns resemble Chianina. It has no phenotypically visible dilution in its coat colour. The likelihood for each trait was 1/4 for Sayaguesa or Chianina traits and 1/2 for Heck traits. 

There is also a (Sayaguesa x Heck) x (Grey cattle x Watussi) bull called Rimu in Hortobagy. It turned out quite nice as you see on the photos. It was a lucky coincidence that it looks this way, it could have also gotten the semi-dominant dilution of Steppe cattle, the short face of Heck cattle or upright horns. The likelihood would have been 1/4 for each of the traits of the founding breeds. 

So when I write “Sayaguesa and Chianina is a good combination” (f.e. here), I do not mean that I believe that F2 will perfectly unite the good traits of the founding individuals in one, but simply that they have the potential to do so. A F2 Chianina x Sayaguesa might either have a perfect aurochs colour, good horns, long faces, long legs and large size (the maximum potential for aurochs-likeness that is in this combination), or it might end up completely white or at least diluted, having tiny horns and being not as large and long-legged as Chianina (so none of the desired traits of the founding breeds united in one individual) – most individuals will end up somewhere in between. F2 will show the full spectrum of all possible phenotypes, no matter which breeds are involved. You can say the same of all possible genotypes as long as you are not crossing or backcrossing with purebred individuals (basically, the un-stable inheritance of Heck cattle, which have a heterogeneous genotype, represents the same phenomenon).  

Thus, breeding is simply a numbers game, you need either luck or you have the produce large quantities of the combinations and select the best out. As cattle unfortunately have the small litter size of only 1, perfect breeding takes a while.