I am trying to fully understand how ETFs work. For the sake of simplicity, let us restrict ourselves to physical replica ETFs.
Let us suppose that an ETF follows an index which tracks three stocks: Alice Enterprises (A), Bob Industries (B) and Charlie's (C).
We want to invest 1000$. The prices of each individual stock are 100$ (A), 10$ (B) and 1$ (C). The index weights at this moment are 60% (A), 30% (B), 10% (C).
If I am not mistaken, when we buy a fund tracking this same index, the fund manager will automatically buy 600$ of A, 300$ of B and 100$ of C.
Now, if we buy an ETF the underlying process is unclear to me. I understand the concepts of authorised participant and NAV spread. If A increases its value to 110$, authorised shareholders will first withdraw shares of A from the ETF and sell those shares, and then they will buy the ETF, pushing the ETF price upwards. This way the ETF price is automatically regulated and the authorised shareholders earn a profit. Everybody is happy.
In contrast to a fund, when we buy an ETF we are not generally buying ETF shares from the ETF manager. Instead, we are buying ETF shares from another investor. Therefore... as I understand it, no new shares are really bought. And thus, following the process above, if we buy the ETF we push its price upwards. This again creates a "gap", this time the price of the ETF is momentarily higher than its underlying assets and, therefore, authorised shareholders will sell ETF shares and buy shares of A, B and C.
So... an ETF manager only buys a fixed amount of shares when the ETF is created?
I want to know if I am correct in this reasoning about how the basics of physical ETFs work. I have tried to keep things as simple as possible. I understand that there will be lots of quirks and quincks, and it is great if these are elaborated in the answers.