Good luck, man. I already did that, and it really paid off. I did it mostly alone, with my brother as musician, with no investment. But it isn't easy. I gone for ensuring success, which meant making a complicated piece of software that is unique and for which i can be certain there is some market without having to second-guess the audience or compete with anyone.
Going the path well walked, gets you competing with all the 99.9% remakes that are entirely invisible. The most profitable game may be cow clicker, but there's thousands upon thousands of other cow clickers that are not profitable, and which you don't see. When you say you look at 'success rate' of X, don't be looking at the % of the successes which are X, that's your basic application of rationality to this decision right here. The simple toy games are not even exercises in game programming. They are exercises in marketing (and blind luck). If you had 10 000 hours of marketing experience, then you totally should go for them.
That's a very good strategy. If I wanted to make my own company, and keep it around for a long time, I would probably do something very similar. Start small and grow sustainably.
You're absolutely right about the cow clicker games. I think their business model has been validated to be used, but the overall game design needs to be refreshed.