What my math suggests at the moment is that keeping crit and mastery at the exact same percentage is what would net you the highest theoretical output because once one gets above the other the lower one will be increasing your theoretical output quicker.
My hypothesis is the same. Crit = Mastery will net you the "best" theoretical returns. It's skirting the line between wasted absorbs and massive overheals.
It's a little more complex math wise for me though, since they are unequal investments. You have to pay more to get the same returns out of crit, but it synergizes with mastery, so it's just a question of where the lines actually intersect.
Gems net the biggest reward per investment from mastery. It's cheaper to buy them. So therefore it makes the most sense to lean that way just from a cost vs reward perspective. Reforging however is a balancing act, so we can play a little more loosely (keeping in mind that mastery is still worth more point for point).
Following that line of thought.... We want gear that has massive Crit (rather than mastery or haste) already on it already. That way when we reforge we're rewarded more just from investment ratio. Better reard from converting an expensive stat to a cheaper one, since you get a 1:1 conversion.
Using Mender's Battletags as an example 637 crit, I can reforge 254 Mastery and get .64% mastery out of it. If I had a similar item with 637 mastery and wanted to do the reverse, I would only get .42% crit.
In my personal experience, I feel that crit is far superior to mastery in 10-man raiding scenes and mastery to be the go to option for 25-man raiding.
With the new mastery scaling in 5.3, crit will become more viable for 25-man raiding.
With 10 man being less forgiving (and 25 mans having more support healing), overheals become less of an issue. I can definitely see how that can tip the balance.