> the insurer won't rate your premium at the cost of risk X; it will simply exclude risk X from your policy and rate your premium

Which makes it odd that there are people who can't get insurance at all because of one factor like a tree or a roof. Instead of these companies (rightly or not) excluding those risks they're just dropping the customers or refusing to insure them.

> Problem is, you cannot 100% predict the future, and there is no model capable of doing so, nor will there likely ever be.

As insurance companies get more data about you from constant surveillance and data brokers it's possible for them to assume things with far more certainty. Worse, they don't seem to care too much about accuracy either. Your health insurance could cost you more next year because data shows more people in your zip code are spending more time in fast food drive thru lanes. People have had their DNA leaked! Predicting the future (accurately or not) is getting easier every day.

> This already happens, incidentally; travel insurance policies will exclude pre-existing conditions or recurrences of previous illnesses as a matter of course

I think that's reasonable for things that nearly certain to happen. It's easy to exclude cancer in a travel insurance policy and still provide some value. It's a lot less likely when it comes to something like flood insurance where your house either floods or it doesn't, although there are certainly houses in places that flood so regularly that they shouldn't be insurable at all and no one should live there. There's a balance that's difficult to strike because the incentive is for insurers to drop anyone who has any real risk.