Chestnut - “red” - is the only recessive in this context. Black is dominant over red every time.
Chestnut is a “recessive” color because it takes the horse being homozygous for the red gene, e, to be red-based.
Black color isn’t recessive - it’s just less common because it takes the horse being Ee or EE (hetero or homozygous black) AND being aa for Agouti.
If B=bay and b=black and c=chestnut, it seems to me the grandparents who produced a black and a chestnut would have to be genetically: Bay stallion Bc & black mare bc
and your black mare would then be bc. (Brother is cc.)
Cross a bc with a chestnut cc and your chances are 50% black and 50% chestnut.
It doesn’t work like that because you don’t have the right breakdown. Red and black are controlled by a single gene - Extension. Each horse gets 1 copy from each parent: It’s either E or e. E means the horse is black-base (not necessarily black). e means there’s a chance the horse can be red-based, but it depends on the other copy: ee would be red-based, Ee would be black-based.
Bay comes from a separate gene - Agouti. A (bay) or At (brown) takes an E? horse and makes him bay or brown. It has no effect on an ee (red-based) horse.
I actually don’t think genetic testing could tell you anything more than bay + black = chestnut & black does.
Huh? Of course it can. Genetic testing tells you everything about the black/red status, brown/bay status, silver, dun, gray, roan, cream, and more.
We absolutely know some things from the above equation, which is that both parents are Ee - they are both black-based (bay and black) so each have an E, and to produce a chestnut they both have to have an e as well - makes both of them Ee.
But since chestnut does not express any form of Agouti, we cannot know the Agouti status of either the bay parent or the chestnut foal, other than the chestnut foal cannot be homozygous for Agouti (because the black parent only had “a” to give, therefore the chestnut foal from that cross can only be Aa or aa).
But if you test that chestnut foal, you’d know whether he’s Aa or aa, and if he’s Aa, if you test at Pet DNA, you’d know if he’s Aa or Ata.
The black foal can only be aa - don’t need any testing to know that. But because he’s by parents who are known Ee, we don’t know if he’s Ee or EE, and genetic testing absolutely easily tells us which he is.
So yes, genetic testing tells us a LOT more than we might otherwise know from phenotype or the parents’ phenotype or even the parents’ genotype
Kind of weird that there weren’t any bay foals (or were there?)
By/out of who? I got lost on this LOL
And I can’t think of any genetic way that you could have 33% odds of any result.
I’m fairly certain there is, since the remainder is 67% which can easily be accounted for by all sorts of variables that we haven’t even touched on - splash, tobiano, silver, cream, etc. You can get % down to a sub-1% possibility. Go to the and put in on parent being perlino, one being buckskin, unknown Agouti and Extension status, both being tobiano/splash/sabino, and have fun with all the potential combinations 
