[Rainymood]

Interesting. It is well known in the high-frequency literature that at frequencies higher than 5-minute one obtains market microstructure noise. The observations you observe do not fully reflect the "true" price. I.e. you observe bid/ask quotes with a spread and the "true" price is somewhere in between. This is due to the bid-ask bounce and other latency factors. How can markets ever be efficient if we can not observe the true price of something?

[totalZero]

There's a transaction cost in either direction.

If you buy a call option, the market-maker buys stock to hedge it. That moves the underlier, and raises the price of the call option he just sold. The reverse is true if you sell a call (or buy a put).

Not only that, but there is a cost for all the people and computers that your order touches as it gets executed.

Without price impact from transactions, markets can't be efficient, because new information has to get priced into the market somehow.

[fabatka]

I'm not familiar with high-frequency literature, but you can rarely measure something with 100% accuracy. Despite this, your house got built even if its measurements were taken with a +-0.5cm error. So if the noise is sufficiently small, you can know the true price with enough precision. I don't know if that is the case though.