Essays on stock return forecasting, trend-following trading strategy and empirical asset pricing

Abstract

The first two essays in this thesis discuss stock return forecast (prediction), a thrilling endeavor of both practitioners and academics of finance with a long history. The practitioners forecast the stock return in real-time to optimize asset allocation and seek an alpha return. In the meantime, recognizing the underlying reason of return predictability may help academic researchers identify what variables explain/drive the stock returns, and thus help them produce improved asset pricing theory. Most of the existing literature on stock return prediction focus on the macroeconomic variables, including the dividend-price ratio, inflation rate, interest rate, volatility, et cetera (e.g., Campbell & Thompson 2008; Welch & Goyal 2008). However, little attention has been paid to the technical indicator (technical analysis) which is extensively used by practitioners (Burghardt & Walls 2011; Covel 2009; Lo & Hasanhodzic 2010, 2011; Menkhoff 2010; Park & Irwin 2007; Schwager 2012). Meanwhile, most of the literature on technical indicator exclusively investigate the profitability but do not investigate the ability of technical indicator in directly predicting the equity risk premium, while predicting equity premium is the focus of vast literature on macroeconomic variables. The only exception is Neely et al. (2014) and they find that technical indicator provides vast complementary information to macroeconomic variables in predicting equity risk premium in the U.S. The first essay extends the playground to China, and investigates the predictability of technical indicator together with macroeconomic variables in China. We choose China for several reasons. Firstly, the Chinese stock market hase become increasingly relevant to not only the academics but also the investment industry. Since 2015, Shanghai and Shenzhen stock exchange together has become the second largest stock market by market capitalization (the largest is NYSE). Secondly, a high level of information friction due to non-transparency and short-sell restriction, and the prevalence of individual investors causing more server behaviour biases (underreaction and overreaction) can boost the predictive power of technical indicators. Lastly, no study has examined the predictability of technical analysis in China, so my first essay filled the gap. We find that technical indicators outperform macroeconomic variables in China and capture ample complementary information. We also find that weekly-level technical indicators outperform monthly-level ones, implying a short-term trending feature of the Chinese stock market. The second essay shifts the focus to the U.S. and other international markets, and is the first study to investigate the predictability of technical indicator in a cross-sectional view. We find that the predictive power of intermediate-term technical indicator identified by Neely et al. (2014) is only useful in predicting the top 10% U.S. companies by market cap, it appears to be a calendar effect, and it does not work well in many other countries. In contrast, the short-term technical indicator can well predict much more U.S. companies, it is not a calendar phenomenon, and it can well predict Japan and other Asia-pacific markets. Finally, contradict to the vast literature on the profitability of technical analysis, we find no positive correlation between volatility and the performance of technical indicators. On the foundation of the Fama and French (2015) five-factor asset pricing model, the third essay proposes three additional risk factors in China based on: 1.) substantial daily-level short-term reversal; 2.) state ownership; 3.) institutional ownership, all of which are unique features of the Chinese stock market. We identify vast useful information provided by our proposed factors and we suggest that the five-factor asset pricing model is not a complete description of expected return in the Chinese stock market.