Adaptive Markets Hypothesis: An Empirical Study Based On China’s Capital Markets

As the basis of classical financial theory, the efficient market hypothesis has gone through nearly 50 years of history since Fama(1965) proposed and has experienced the rise of the 1970 s,peaked during the 1980 s and after that its influence gradually weakened in the 1990 s. While behavioral finance which is opposition to efficient market hypothesis has taken off from the 1980 s and become prosperous in the 1990 s, then recede in the early 2000 s. The controversy between these two schools promotes the development of modern finance together. Undering the rise and fall of different market theory are huge changes in the global financial markets, compared with decades ago, the increasingly complex financial markets faces higher uncertainty. Since the Asian financial crisis, the United States subprime mortgage crisis, the European debt crisis, the traditional market theory including the efficient market hypothesis and behavioral finance has been powerless in face of the real financial world which like a roller coaster and can’t give a reasonable explanation. There is a pressing need to development new theory to explain the dramatic changes in the financial markets.Lo’s(20004,2005) adaptive markets hypothesis(AMH) has attracted great attention in the learned world gradually. The primary components of the AMH consists of the following ideas: ①Individuals act in their ownself interest; ②Individuals make mistakes; ③Individuals learn and adapt; ④Competition drives adaptation and innovation; ⑤Natural selection shapes market ecology; ⑥Evolution determines market dynamics. Scholars adopt various methods to test adaptive market hypothesis and achieve some research results. The traditional financial theory have not yet acclimated to the new environment because of China’s specific national situations and multidimensional characteristics in the capital markets, this reality also reflects the limitations and shortcomings of the traditional financial theory. Some domestic scholars have noted the AMH prospectively and try to expatiate on the theory, but the empirical analysis on AMH combining China’s capital markets is rare. In this case, the use of financial data for empirical research on AMH becomes very valuable.Based on the Status of lacking of quantitative research for AMH test, this paper test the AMH from four perspectives which contains the dynamic market efficiency, trading strategies evolving, investors learning and competing, time-varying risk premia to verify whether AMH be able to explain China’s financial markets. This article will study from four aspects mainly.Firstly, this article tests the dynamic market efficiency. We empirical research the return predictability of Chinese stock market using the modified automatic portmanteau Box–Pierce test,wild boot-strapped automatic variance ratio test and generalized spectral test apply the daily and weekly data of Shanghai Composite Index and Shenzhen Stock Index from 1990 to 2013.At the same time we test the time varying nature of return predictability via rolling sub-sample window.The study finds that return predictability is driven by changing market conditions.Returns are found to be unpredictable most of the sample periods,some short-lived statistically significant return predictability can be associated with major exogenous events such as Finance Crisis.The results are in support of the adaptive markets hypothesis,which claim that changing market conditions drive the key market features such as the return predictability.Secondly, this article test the evolution of trading strategy with program trading method.The adaptive markets hypothesis posits that trading strategies performance will evolve as traders adapting their behavior to changing circumstances.This paper studies the performance of ten different technical trading strategies base on Chinese commodity futures to test whether the result accord with the adaptive markets hypothesis.The empirical results show that the average returns of simple and known strategies such as MACD、RSI、BOLL、MA+MACD are negative,while the average returns of complex and uncommon strategies such as MA portfolio、KD、SAR、ADX+KD、MA(5,20)、ATR are positive significantly,further more,the performance of trading strategies periodically fluctuate as price trend. The test confirms that the trading rule excess returns still exist in Chinese commodity futures markets and the evolution rule is consistent with the adaptive markets hypothesis.Again, the papers adopt machine learning algorithms to predict and trading of stock index futures to emphasize the importance of investors learning, competing and innovating indirectly. According the concept of machine learning algorithms, this papers propose a combined forecasting method based on support vector machine and random forest, and we conduct an empirical study for the stock price index future return from 2010 to 2013. Firstly, we rank the importance of stock index return effecting factors using random forest, and then we attempt to predict the direction of stock price index movement with random forest and support vector machine, at the same time we compare the predicted results with back propagation neural networks,finally we carry on simulated trading. Other things being equal, empirical experimentation suggests that the trading strategy based on random forest has good performance, Investors can obtain excess returns by learning and innovating trading technology. AMH can explain the research conclusion.Finally, kalman filtering and smoothing models are adopted to empirical research the variant beta of five style index. AMH argues that systemic risk of financial assets varies with the market environment, the relationship between risk and return is not linear, investors will take adaptive investment strategy depending on the different market environment. This paper use kalman filtering and smoothing models to test the variant beta feature of five representative style index and analysis the reasons for beta changes with event study method. The empirical results show that the beta of the style index is time-varying, and associated with some significant economic, political events. The classic CAPM model can not explain these phenomena, while AMH can explain better.