Anyone knows how fingerprinting works mathematically? what do they use? when the signals may vary even slightly, your final hash should be sufficiently accurate and stable. A simple hash won't do it.

Changing IP, missing cookies/session, resizing window, TLS/SSL browser handshake, screen depth, font rendering, MTR, network latency if you want to measure, etc... any signal alone means little but as a whole mathematically what is it possible to use to get a pretty good stable hash? something I want to learn more about. I guess it has something to do with vectors and clustering signals but no idea and would like to know what exists mathematically to archive this (if possible or with very small error).

And then I want a browser that lets me opt-out all WebAPIs ever possible.