This thesis emphasizes on the broader implications of market frictions on the beliefs and preferences of investors. Model-free frameworks are proposed to infer such implications using information conveyed by asset prices and survey data. While Chapter 1 examines the expectations and beliefs of investors, Chapters 2 and 3 focus on investors’ preferences, or, more generally, Stochastic Discount Factors (SDFs). In particular, Chapter 2 studies SDFs of international investors with the objective of identifying a global factor structure of exchange rates. The model-free frameworks provided in this thesis give the means to rationalize the use of regularization techniques, common in the Machine Learning literature, for portfolio and SDF problems. The chapters demonstrate that the trading frictions that investors experience when forming portfolios can serve as motivation for such regularization techniques.