Same way the best engineers "prepare" for a job interview - by largely ignoring it. Instead of studying Java certification books they hack in Python, C, and Haskell and read about interesting algorithms in their spare time. When it's time for a job interview, all they have to do is show up.

If you talk to one of your target customers and one startup founder in a similar field a day, in three months you will have spoken to ~180 people that are intimately related to your business. You'll know more about your domain than you ever wanted to, and when the YC interview comes, there is nothing to prepare for.

That's the whole point it's not about "preparing and practicing" for the pitch/investors. You'll never "know enough". Instead if you're passionate about the domain and somewhat intelligent then you should know everything the investors know and a hell of a lot more. The investors should have a fraction of the knowledge and experience that you have and thus should be practically incapable of asking you a question you don't know the answer to.

If you can't answer the question then you're either not particularly passionate, or you're not very smart - either way you're not worth investing in.

I don't know if that's true. Some ideas anyone can have a cursory knowledge of your domain, and can identify corner cases you may not have considered. That's mostly the stuff I'm concerned about. I'm curious what kind of corner cases people can ask questions about.

Then you tell them you don't know and you walk them through an educated guess. They just want to know that you understand your supposed area of expertise.

(full disclosure: haven't pitched to investors/YC)

Just talking about your product with colleagues, mentors, friends, etc. can help you get prepared. Most likely they will ask you questions and it can be like a mock "interview".