If you are taking the hypothesis route, you start with a prior (e.g. Trumps tweets have a bearish effect on the market) and establish it's statistical significance. After that step you already know if you have alpha and have some clue about how to trade it.With regards to backtesting, the key is how much predictive power does your set of parameters/variables have.
It does not mean that every idea that has alpha will end up as a trading strategy either, since you can find that's hard to monetize. The parameters start coming in once you embark on transforming your hypothesis into a trading strategy. If you know your "kraft", these parameters are very benign and your should have fairly stable response curves to them.
I have no idea who he is, to be honest, but he does not seem to have a Bloomberg accountI think Robert Pardo is the real deal, it's just that he suffered a big drawdown around the 2008 timeframe, and that's probably why he published the 2nd edition of his book in 2009.