We develop a new approach for testing conditional asset pricing models by asking what realized returns we would observe, given the pricing model under scrutiny. This avoids the issues in usingrealized returns as a proxy for expected returns. The new approach is used to test the Merton (1973, 1980) model and a long-standing risk-return puzzle: the price of market risk has often turned out to be negative and insigni cant. Comparing the price of market risk estimates from the new and the traditional testing approaches, our results on US data give strong support for the new approach.