I'm not entirely sure if this thread is the place to discuss the RogueThing tool or not, or even if someone who can answer my question reads the EJ forums to begin with, but I'll try none the less I suppose.
I originally reforged my gear based on just a base knowledge of the class mechanics and how I felt to best optimize the total dps gain of reforging within those mechanics: try to nail expertise on the exact rating I wanted, spell hit on the exact rating I wanted, best stats past that point when applicable, reforge pieces with our best non-capping stat (haste) on them when possible so that later pieces can go into haste, etc. In doing so I ended up with a piece which appeared to be optimal not to reforge at all (hit and haste with the hit being entirely spell hit). After coming across the RogueThing tool and adjusting weights and boss dodge, I checked out the auto-reforge tool to see if what I came up with was optimal or not. Interestingly enough, the tool recommended something fairly substantially different. So I jotted down the total EP readout of the suggested reforging and then plugged my own guess into the tool, only to find that my rough eyeball of the best reforge gained me 33 EP over the suggested one.
So with that said, I'm pretty curious as to what sort of methodology the auto-reforge routine takes, and possibly even more so, what the truly best method is (aside from brute-forcing every combination of course).
To outline my method, if it helps spur some ideas in others, I first listed every numerical value of stats that I could confidently gain (i.e. dumping crit) on every piece except for my mentioned piece
which had hit/haste. With these values all listed, I went through a trial and error process until I found combinations that produced exactly 172 expertise. With these values crossed off the list, I went through another trial process which brought me as close as possible to the spell hit cap (I think I ended up 3 over). From here the rest was fairly straight-forward, just reforge crit into whatever stat was left that was the best.
Now I'm not saying my choice was the absolute 100% best, and in fact I'm fairly sure it's probably not, given the restrictions on reforging (must pick a stat non-present) and the sheet number of possible combinations, but I do find it at least interesting that my crude guess beat the reforging routine in the tool by 33 EP.
So, with all the exposition out of the way, how does this tool produce its result, and more importantly, how should
it produce it's result? If reforging could add any stat regardless of what the item started with, I think the optimal path would be fairly close to what I outlined in my thought process, but as long as we have that restriction to work around, I think the truly best path is a bit more complicated than people realize.