Performance Evaluation of Trading Strategies: A Regression-Based Approach(2/3)

In the past few decades, many asset-pricing anomalies, which often appear in theform of long-short strategies, have been documented in the literature. To serve asa piece of evidence against the efficient market hypothesis, an anomaly should beproofed that it survives under proper risk adjustments, and that its return patternsgo beyond those of the existing anomalies.The price momentum effect proposed by Jegadeesh and Titman (1993) is perhapsthe most often-cited among the many anomalies. However, there are a couple ofproblems with their strategy. First, it is equal-weighted both over time and in crosssection,which requires constant rebalancing on a monthly basis. Second, it failsto account for microstructural biases such as nonsynchronous trading and bid-askbounce. Third, the performance is calculated on simple average returns, withoutconsidering the compounded nature of real-world returns. As a result, the momentumperformance is overstated.As the problems related to JT’s price momentum effect are also common for allother anomalies, it is important to come up with a research method that allows forexamining both the statistical and economic significance of the anomalies. To tacklethis research question, I shall propose a regression-based framework initially proposedby George and Hwang (2004), with some significant improvement. The proposedmethod would be able to resolve the problems associated with JT’s strategy, andallow for examining the robustness of competing strategies (or effects) from bothrational and behavioral perspectives.The project has two major objectives:1. I will provide a unified regression framework that allows for examining the profitabilityof various competing strategies and their driving force. I will conductan empirical analysis using the U.S. data, and perhaps would also conduct asimulation analysis to examine the extent to which how various confoundingfactors affect the performance of trading strategies.2. I will conduct two empirical applications based on the proposed method.(a) Investigate the competing profitability of three trading strategies (price momentum, 52-week high momentum, and five-year low) and their drivingforces.(b) Examine the sources of various anomalies (e.g., Stambaugh et al (2012)and Novy-Marx (2013).5