One of my biggest takeaways was learning about "crackmes" - which are small challenge binaries designed to be reverse engineered in order to learn the craft. They're kinda like practice locks in the lockpicking community. The book comes with a bunch on a CD-ROM from memory - but there's plenty more online if you go looking. Actually doing exercises like this is the way to learn.
You don't start trying to reverse engineer COD. You build up to it.
greetz to readers of Unknowncheats, cs.rin.ru, etc.
Milworm (milw0rm?) also got me started back in the day.
UnknownCheats is also absolutely amazing for cheat development. Back when I was writing undetected kernel cheats for my own experimentation purposes, I learned so much there.
UnknownCheats was (still is?) good for getting information on undocumented APIs when game modding (for a good while the Half-Life SDK was incomplete).
Its a hard first step, but I highly suggest you take the time to analyze a small binary, starting with understanding the registers for the architecture, understanding the different function calls, and then looking at the elf file and analyzing every section and how static linked libraries work, and how dynamic linking works with PLT/GOT. GPT models are REALLY good at helping you understand this, and you can also use Ghidra for decompilation. Do everything on Linux btw, as the tools are very easy to use and much less Cumbersome than windows.
Once you understand all of that, tracing assembly is pretty easy - its either register move operations, math operations, compare operations, jumps, and function call and returns (which basically are just shortcuts for handling the stack frames), with a few special instructions here and there which are usually just some optimizations that you can look it up ad hoc. Once you get handy at ghidra, you can look at decompiled C code and start replacing variable names to make the code readable, and then you generally get a good idea of project flow.