Unless you're going to be wildly successful the moment you launch, you're better off picking one stack that you understand, creating a working version, then iterating to improve on it. If you try to do it perfectly or make it as fast as possible from the start you may never finish - and you'll never be able to test your idea.