Estimating likely changes in the future based on current information is different from divining the outcome of a completely independent chance-based event.

If I can buy X for $1.00 from location A and sell it for $1.01 at location B, then the only thing I need to worry about is what happens if the price moves against me at either A or B between the time I send my orders and the time they're accepted.

I guess you could technically call that "trying to read the future", but the (big) difference is that the horizon is on the order of probably a few milliseconds at most, not 5 minutes.

If it's a temporarily biased coin, you can make some money off of it.

If it's a fair coin you can make money off it, and convince yourself you can predict the "market" "because the coin is biased" ... but it's just 'luck' (ie statistical anomaly).

It evens out in the long run so this doesn't apply to this person's trading.

I think that's a statistical misunderstanding. It's likely to even out; across a population it will even out. An individual however can get "more heads than tails" by chance.

People can read five minutes in to the future.

Five minutes from now I'm going to be sitting on my bed.

Five minutes from now I'm going to be male.

Five minutes from now I'm going to have a nose.

That I can read these aspects of the future has no bearing on me being able to tell the future of coin flips.