This project investigates the short-term to medium-term predictability of stock market excess returns from an empirical point of view. Empirical evidence on predictability of market excess returns implies that risk premia are not constant but varying over time. The importance of the topic can be seen not only by the amount of empirical literature but also from the relevance for theoretical financial markets research: A growing literature now seeks to match the empirical finding of stock return predictability in theoretical models. Besides conducting classical predictability tests for the German stock market, the project also has a special focus on the study of expectations data from the ZEW Financial market survey and their explanatory power for future stock excess returns.