[tex]\frac{Number \left \left of \left \left outcomes \left \left with \left \left a \left \left particular \left \left phenotype/genotype}{Total \left \left number \left \left of \left \left outcomes}[/tex]
For example, if you performed Aa x Aa, then there are four possible outcomes, AA, Aa, aA and aa. These are all equally likely, and two are the same. So, for AA, it appears once as an outcome of four possible outcomes, thus you get 1/4. For, Aa (aA is the same), is appears twice out of four possible outcomes, thus 2/4, or 1/2. For aa, it appears once out of four possible outcomes, thus 1/4.
I think that's what you were looking for. For crosses where a large number of genes is being considered, punnett squares and writing out the possibilities can be tiresome; and just using probabilities is easier (although possibly more prone to error).