I have not seen evidence of fake traffic at Huobi, but I suspect that there may be fake traffic at OKCoin.
First, traffic at OKCoin slows down at night, but does not stop completely like at Huobi; there is a steady backgrount traffic 24/7.
Second, most Huobi transactions occur randomly at the two ends of the current spread, as one would expect from buyers and sellers independently moving to meet each other's position; whereas many of those at OKCoin seem to occur in at random places in the middle of the real spread, as if the buyer and seller had a rendez-vous there.
OKCoin also seems to have an excess of small transactions. (I should make a histogram to confirm that.)
Here is a plot of a few minutes's worth of transactions. Each dot is a transaction (except that transactions with same timestamp and price may hide each other). Dot size is proportional to BTC volume.
Am I seeing things?
I'm seeing dinosaurs, too..... or maybe train cars being assembled in the form of a carrousel? Kind of like a circus.
I do NOT claim to know very much about how one exchange may differ from another or the way that a bot may work based on the way it interfaces with the exchange. However, couldn't some of the explanation of differences be based on how bots may need to be programmed differently based on the exchange rules, fees and maybe based on times of greater or lesser volatility?
But, yes, I can understand why you may consider certain patterns to infer that the trades may NOT really be taking place b/c they do NOT make sense to be occurring under those circumstances... even if it were due to the programing parameters of bots... that are paying nearly zero fees.
