You learn valuation (and find out how much you don't know) by valuing businesses and companies, not by talking, reading or ruminating about doing valuation. That said, it is natural to want to value companies with profit-making histories and a well-established business models in mature markets. You will have an easier time building valuation models and you will arrive at more precise estimates of value, but not only will you learn little about valuation in the process, it is also unlikely that you will find immense bargains, because the same qualities that made this company easy to value for you also make it easier to value for others, and more importantly, easier to price.
I believe that your biggest payoff is in valuing companies where there is uncertainty about the future, because that is where people are most likely to abandon valuation first principles and go with the herd. So, if you are a long-term investor interested in finding bargains, my advice to you is to go where it is darkest, where micro and macro uncertainty swirl around every input and where every estimate seems like a stab in the dark. I will not claim that this is easy or comes naturally to anyone, but I have a few coping mechanisms that work for me, which I describe in this paper.