Personally, I've only observed Bitstamp leading and Huobi following it reluctantly. I don't know why some perceive the opposite.
I don't like the idea of thinking of it as "fake" volume, but I do believe it needs to be heavily "discounted" for purposes of volume analysis.
Based on some (admittedly rather crude, please don't ask) method I employ, today's calculation for example would look as follows:
4400 (stamp) + 3000 (btce) + 5400 (finex) = ~13k "Western USD volume"
43000/10 (ok) + 29000/5 (huo) + 3000 (btcchn) = ~13k "Chinese volume"
i.e. today I discount OKCoin volume by a factor of 10, and Huobi by factor 5. It's coincidence that today they add up to roughly the same value, btw.
Note that I don't buy the claim by finex that they really cut *all* ties with stamp, so some of the finex volume is likely to be a duplicate of stamp volume, so total western USD volume is possibly too high. On the other hand, it doesn't include EUR (and other currencies) volume yet.
In summary, on a day like today I consider, at least by volume, the "Chinese" influence about as big as the "Western" influence.
How did you arrive at those percentages that 90% of OKcoin's volume should be discounted and 80% of Huobi's? Seems kinda arbitrary to me. Why not 50% or 30%?