Drastic Econometric Failures and Theoretical Pitfalls in French’s Method of Day-of-the-Week Anomaly Detection

Stock market anomalies, such as calendar effects and price bubbles, are events that cannot be explained by the famous efficient market hypothesis and have always attracted the attention of economists and statisticians. In this regard, in 1980, Professor Kenneth French published a pioneer study on calendar anomalies focusing on the day-of-the-week effect (<xref>French 1980</xref>). The novelty and easy-to-use aspects of his approach made it the benchmark model for studying such events in financial markets all around the world. However, its fame and practicality hide the model’s serious econometrical failures and theoretical pitfalls. The present study attempts to criticize French’s model from a technical point of view and then tries to introduce an alternative approach to studying such phenomena. The advantages of the new method, which is called mean/median analysis, and the failures of French’s model are illustrated through an empirical investigation of the daily yields of US 10-year Treasury bonds.