From Shamaya's post:
So from that we already know that we can't trust his calculations, if he's using a 3.0 max instead of 3.15.*capped pDif multiplier range w/ crit, assumed to be roughly 2.6 to 3, based on public data. (edit: I know the spread is larger than this, and I know the max is slightly higher than this. I just don't know how high.)
Sample size is small. 8 baseline, 3 at each str value (with 1 point difference in str), 3 different dex values (6 of one value, 1-2 of the others)
Sample range is generally about 10 points shy of the max possible spread, which introduces huge margins of error with the way he's doing his calculations.
There are no data points with a gorget, so we have no simple means of calculating fTP. He back-calculated fTP based on WSC, which is a flawed approach. It may be possible, but not in any way using the math he showed.
In any case, I don't need Byrth's min/max values to be the 'true' min/max possible. I verify my analysis primarily on the 3.0 marker, which is easily distinguishable with a sufficient number of sample points. Byrth didn't quite reach 'sufficient', but he provided enough that I could narrow things down to a pretty small range.
However, for the sake of argument, suppose we choose another fTP value. We'll consider 1.3.
1.3 applied to the baseline requires
WSC percentages of Str and Dex to generate a total of 45-46 from 79 str and 84 dex. (note: 46 from baseline, 45 from +0.2 fTP)
WSC percentages of Str and Dex to generate a total of 62 from 79 str and 140 dex.
WSC percentages of Str and Dex to generate a total of 63 from 131 str and 95 dex.
From the latter two, 52 str has to be worth 1 more point than 45 dex. At the same time, 56 dex has to be worth 16-17 points. So 45 dex has to we worth 12.86 to 13.66 points, and 52 str has to be worth 13.86 to 14.66 points.
Scaling down the differentials to per-point and applying the alpha value, dex would be worth 33.6%-35.7% and str would be worth 31.4%-33.2%. So even with a different fTP, str is still about 2% below dex in WSC.
Now reapply those to the baseline: 21+24=45; 22+25=47. Probably somewhere in the middle, like 34% dex/32% str (45 total), or maybe 35% dex/32.5% str (46 total).
So yes, it's entirely possible to come up with different WSCs that match a given fTP value. The problem is determining the fTP in the first place. Adding 0.2 fTP to a 1.3 fTP weaponskill will gain you 2% more damage than adding 0.2 fTP to a 1.5 fTP weaponskill. You have to be certain you can distinguish that difference.
Interestingly, it would be easier to distinguish with higher base damage. The typical methodology is to minimize stats for the baseline, which is used to get the fTP, then add stats on top of that to determine WSC. It may be better to max out the stats when calculating fTP, then remove them to figure out the WSC (WSC works solely on the difference in the stat, so moving up or down is just a matter of preference).
In any case, in my original post on this I went through quite a bit trying to work out the proper fTP to use, and settled on 1.4. Unfortunately there aren't enough data points to be absolutely certain of it since Byrth stopped including all data points and only recorded the lows and highs. 1.4 simply seemed the best overall fit. If you're willing to provide me with 30-40 sample points each of with and without gorget/elemental belt, I could be far more certain of my analysis.