A variety of variables have been used to form contrarian portfolios, ranging from relatively simple measures, like book-to-market, cash flow-to-price, earnings-to-price and past returns, to more sophisticated measures based on the Ohlson model and residual income model (RIM). This paper investigates whether: (i) contrarian strategies based on RIM perform better or worse than those based on the Ohlson model; (ii) contrarian strategies based on more sophisticated valuation models (e.g. Ohlson and RIM) perform much better than the relatively simpler ranking variables that have been used so extensively in the finance literature. Given that the RIM and Ohlson models require greater information inputs and technical know-how, and make different implicit assumptions on future abnormal earnings, it is important to ascertain if they offer significantly greater contrarian profits to outweigh the increased costs that they entail. Indeed, our surprising finding is that simple cash flow-to-price measures appear to do almost as well as the more sophisticated alternatives. One would have expected the sophisticated models to significantly outperform the simple cash flow to price model for the reasons given by Penman (2007).