It depends what you mean by "risk free" and it depends what you mean by "out-modelling" and neither of those points actually have any information about what making money involves. It is extremely difficult to make generalizations but in an extremely general way it is about working out the price of x in terms of y and z. So you price security x based on security y or time series z. Clearly, that is never going to be risk-free.
But even that is a massive oversimplification nowadays as people are using machine learning and the like which is probablly completely zero-sum in the long-run i.e. you only make money if no1 is doing what you are doing. Imo, it just isn't worth getting involved with unless you have a serious academic background in a related field (machine learning/speech recognition/etc.) and have the "resources". Having said that, given the stuff that people do on here it is probablly a significant improvement. Either way, the stuff in Wilmott is, it seems to me, more for people looking to work in IBs where the person facing you is extremely stupid or has bad information or you want to work in risk.
