Sure, in an idealized enough world you can make many statements true. In the real world people buy and sell stock for many reasons, and value different types of returns on different time horizons, and it’s likely that for a majority of stock trades both the buyer and seller are happy with the trade.

People have different risk exposures and different risk appetites. If you're a corn farmer then your entire livelihood is tied to the price of corn. You're willing to give up expected value to reduce risk by selling futures and there are others who are less exposed to corn prices who are willing to take the other side of that trade.

In your theory, there is no cost to servicing loans (on both sides), foreclosing on an asset or long term lockup of capital in not perfectly liquid markets. Or other uses for shares (eg exerting corporate control) absent receiving dividend streams.

With the exception of corporate control, this is covered in my final sentence.

None of my costs had to do with middlemen.

Loan servicing costs only exist because of middlemen. If two people directly interface for a loan there is no additional cost (other than the cost of transportation, which only exists because a middleman is charging for gas and car maintenance).

Foreclosing on an asset doesn't happen in my example is not a cost, but a risk, which the interest rate should be accounting for.

Longterm lockup of capital is not a problem, as either party in my example are financially gaining regardless.

And I granted you the other uses for shares, which technically can be done without buying the share, but at a middleman cost (renting the share).

I really don't care to continue this. Mine was just a hypothetical in favor of the idea that there's always a "loser" (when compared to all possible financial alternatives) in stock trades, even if everyone ends up happy with the trade. Sure, there are holes in the hypothetical. I don't care.