Grey - Grey is technically considered to be a white marking, and is a modifier that affects any and all coat colours, turning them grey if the gene is present.  Again, with grey, it does not matter if there is a single gene or both grey genes present, the horse will be equally grey.  Greys are born their base colour and grey as they age.  Grey can be represented by "G" and "g".  Grey works over all other modifiers as well, so a horse could be reb based with  cream and dun dilutions, but it would still grey out as it matured.  Since grey is so dominant, it does not hide or skip generations, a grey horse always has at least one grey parent.

Roan - Roan, like grey is a type of white marking a foal may be born roan or it may develop over time, although unlike grey, it does not overtake the entire coat, hairs of the base colour and white will be mixed.  Some roans are born solid, and develop white hairs throughout their coats except on their heads and lower legs, which remain dark.  Roaning can also affect any base coat colour.  Roan is another dominant gene, so only one copy is needed to cause a horse to roan.  It can be represented by "R" and "r".  Roan turns a black to a blue roan, bay to bay roan, red to red roan etc. "Roaning" is a term also used to describe smaller areas white and darker hairs mixed on a horse.  The inheritance of this incomplete roaning is not fully understood. 

Other Modifiers - There are many other modifiers, including the paint and appaloosa patterns, various marking patterns like sabino and rabicano, and the Champagne dilution, which I have not touched on here. 

To determine percentages of colour possibilities in offspring, we first start with the base colour pair as in the first punit square example.  After those percentages are determined, you can add in other factors for the horses in question.

At left, we see that the base colour gives us 50% black (Ee) and 50% red (ee).  The second one tells us that 50% of those base coloured foals would receive the cream gene.  That gives us 25% Black (Ee cc), 25% Smokey Black (Ee Cc), 25% Palomino(ee Cc), & 25% Chestnut (ee cc).

When we add a 3rd variable like agouti, then we have to break down each of the possibilities we have in half again.  Since the mare shows no signs of brown areas, we assume she has no agouti, or is "aa". The stallion is red based, so could possibly carry at least one hidden agouti gene.  Let's assume he is heterozygous for agouti (Aa).  If you do the punit square, you will see that again, that would give us a 50% chance of the foal receiving a copy.  So, that would give us...

12.5% Black (Ee cc aa)

12.5% Bay (Ee cc Aa),

12.5% Smokey Black (Ee Cc aa) 

12.5% Buckskin (Ee Cc Aa)

12.5% Palomino (ee Cc aa)

12.5% Palomino(ee Cc Aa)

12.5% Chestnut (ee cc aa)

12.5% Chestnut (ee cc Aa).

Although genetically different, the palominos and chestnuts with and without the "A" would look the same, so it would actually be 25% pali and 25% chestnut

Our stallion Lenny is a Sorrel (chestnut), so his base colour is ee.  This means that for base colour, he can only contribute a red gene.  Since red is recessive, that means that in most cases, the mare's colour will be more significant in determining the colour of a foal.  It is unknown if Lenny carries Agouti or not, except that if he carries it at all, it is only one copy since he has sired a black foal.  Even though there are few bays in his pedigree, agouti can be carried through several generations of red based horses.