"Could" doesn't apply after-the-fact (or, possibility arises from the model)

People often speak of things that they think "could" happen. (They may use a different word, like "can" or "possible", but the sentiment is the same.) E.g. "it could rain tomorrow" or "I could just punch the person I was talking to". (Please do not actually punch the people you talk to.)

If the speaker has yet to observe whether the "possible" thing has happened, which usually happens sith the event's "opportunity" is still in the future, "could" has a reasonable meaning. It expresses uncertainty about whether the thing has happened: to your current information, future worlds would seem plausible whether or not the thing happens. In the examples, that situation arises if, say, it hasn't rained in a few days (and you're not in a desert, and you haven't looked at a forecast), or if you're mostly nonviolent but enemies with the person you're talking to.

But after the event either happens or doesn't (the "opportunity" passes), or you otherwise come to know for certain whether or not it will happen, there remains no reason to use "could".

The only distinction between a thing that "could happen", but doesn't, and a thing that "couldn't happen", is whether you consider it in your incompletely-informed understanding of the world. If you think there is another distinction between those cases, I would like to hear about it. No one I asked has been able to answer it in a way that stands up to casual scrutiny.

Once you know whether or not it rained tomorrow (by experiencing tomorrow, or looking at an accurate forecast) — once you know whether or not you punched that person (by finishing your meeting, or mentally committing to not punch them) — your incompletely-informed understanding of the world should only consider one option. When you know what really happened, the only things that "could" happen are the things that do happen. If you see that it did not rain tomorrow, for "tomorrow" is now yesterday, it could not have rained yesterday. If you did not punch that person, you could not have punched that person.

This principle applies equally well to outside events and your own decisions. "Free will" is how it feels to compute expected utility (figure out the best actions), to lead to a decision while you remain uncertain about what you'll do. Speaking in its terms (those of "possibilities" and "options") after you decide is useless.

After you have observed an outside event, or "observed" your own decision, there are reasons you should still be a bit uncertain (the ever-present mild risk of hallucination or false memories). There are causes for the exact events with no measurable basis (quantum randomness). You may face similar decision-questions in the future to the one you went thru just now (such as whether to punch the next person), and consider what to do then based on what you did now. But these uncertainties and "opportunities" work differently from the kind people ordinarily express with "could". The casual phrasing loses the distinction and suggests an incorrect interpretation.