Although I find classification studies with simulated data largely pointless, I'd rather add roughly two simulation configurations to ensure the paper is published. If a referee desires more than two, so be it. But two should be good enough.
After a discussion with Dean Young on the matter, we have decided to forego simulated data. If a reviewer asks for this, then we can provide it with little trouble.
Although I find classification studies with simulated data largely pointless, I'd rather add roughly two simulation configurations to ensure the paper is published. If a referee desires more than two, so be it. But two should be good enough.