Your fee arithmetic is right, and it is the same wall we ran into when deciding
how to grade signal channels — worth sharing because it comes at the question
from a different angle.
When we replay a channel's published signals, the yardstick is whether price
moved 2% in the called direction before the stop was hit. 2% is deliberately
large. At a 0.25% target, a 0.05% taker fee is a fifth of the whole trade, so
the measurement would mostly be measuring the exchange's pricing rather than
anyone's skill. Anything at scalp distance stops being a statement about the
call and becomes a statement about costs.
Two numbers from that dataset that bear on your plan:
- Under that 2% rule the median channel is right about 72% of the time, and 23
of the 29 channels with closed trades are still cumulatively negative on
their own entries and stops — before any fees are applied. High hit rates at
short distances are common and survive contact with reality badly.
- Of 10,400 replayed signals with a verdict, 4,093 never filled at all: price
ran past the entry band or never reached it. For a scalper that is not a
footnote, it is the business. Your model assumes you get the entry; at 0.15%
distances you often get either a worse fill or none.
None of which says scalping cannot work. It says the edge has to be found in
execution and fee tier, because at that distance the fee schedule is a bigger
input than the setup.