That’s a very reductionist summary of what I wrote that I wouldn’t consider accurate.
I disagree that there is nothing to learn from what happened, even without considering alternative reality (by the way, I’m not arguing against that, as I said that’s just closer to fiction than history, but that’s still an interesting tool). You can see patterns and systems, name and describe them, try to see if we can find similarities with other events. But you cannot change their output, because it already happened.
Randomness is defined by the lack of patterns and predictability, in practice a system chaotic enough will be considered random (see your toss of coin, in fact it’s just a physic problem with a very high number of parameters).
I believe that you give way too much credit to the concept of pure chance (that you don’t define). I personally believe that human history is mainly the result of high level systems interacting with each other (thus can be described and analyzed), studying them backward seems to be prone to a lot of bias as we are trying to make them reach what we consider to be the contemporary state.