I'm not 100% sure this is correct but here goes my understanding/theory on decoloration.
Goldfish are tetraploid. Humans have two sets of genes where as goldfish have 4. So if a human couple has two sets each there are four possible copies to choose from, two from each parent. A pair of goldfish with 4 copies each has a total of 8.
So to over simplify you have 8 chances to get the decoloring gene in the offspring. With decoloring the effect is on or off, no partial, just slowly changing. So if a fish starts to change it will ultimately change fully. This gene actually kills the melanin containing cells called melanophores. I think a similar effect happens and kills the xanthophores which house red/yellow pigments.
I think there are environmental things that can effect the speed at which the decoloring happens but once it is set into place it is total. Fish like this:
http://goldfishgarage.blogspot.com/2009/11/large-black-and-gold-common.html are supposed to have some of the late decoloring genetics of the shubunkin but the metalic and decoloring are dominant. I suspect these fish were grown in cold water and had very few copies of the decoloring gene.
If a fish has all four copies of the decoloring gene I suspect they turn very early and visa versa if they only have one copy.
So if you start with orange or white fish the offspring will always eventually turn red and possibly white.
Instead of being the good old punnett square and medals peas goldfish genetics is like playing with a rubics cube.