Rational Speculation and Heavy-tail Phenomena: A Bootstrapped Optimization Approach

• Main Authors: Yuexian Lin, 林岳賢
• Other Authors: 楊朝成
• Format: Others
• Language: en_US
• Published: 2010
• Online Access: http://ndltd.ncl.edu.tw/handle/98785374801169474008

博士 === 臺灣大學 === 財務金融學研究所 === 98 === Traditional wisdom regarding market efficiency claims that there is no hope making trading profits consistently by investigating the price history only. In this monograph, we discover an alternative possibility---the result is against the weak-form efficiency. Methodologically, if a consistently profitable trading rule can be found and is well-adjusted for data-snooping bias, then the weak-form efficiency hypothesis could not be sustained. This is true because such a thing cannot exist in the efficient market world. The positive-feedback trading requires the demand function for a risky asset is increasing with respect to prices and is intimately related with the momentum investing. Nonetheless, we argue that a zero cost portfolio might generate enormous interim losses before the liquidation of positions or realization of momentum profits. Since no one has unlimited wealth, investors who use momentum strategies in the hope of future profits might be forced to exit the market prematurely due to huge interim paper losses. Consequently, the potential benefits from momentum investing must be conceded if investors are wiped out from the market beforehand. This effect is especially significant in futures markets, where the leverage is crucial for participants with wealth constraint. The purpose of this study is twofold. Firstly, originating from the idea of DeLong et. al (1990), we propose a theoretical argument for rational speculation that relates the positive-feedback trading to the heavy-tail phenomena. A simple yet realistic positive-feedback trading system is then constructed to capture heavy-tails in asset returns. We realistically incorporate standard risk control mechanisms and examine the rule via resampling techniques to avoid curve fitting problems. The positive-feedback trading rule has passed the challenges of non-parametric blockwise bootstrapping methods. Extensive numerical experiments suggest that these trading profits might arise from the heavy-tail phenomena, and verse visa. Secondly, there is an open question prevailing in the realm of program trading: why are the actual trading performances, with few exceptions, systematically worse than the historical back-testing results? It turns out that the answer is due to the institutional regularities and is quite straightforward under the perspectives of the bootstrapped optimization scheme. In our opinion, all the trading performance measurements would better be represented in a manner of bootstrapped distributions, rather than based on a single sample path. We hope our investigation for rational speculation and quantitative trading provides helpful insights for both practitioners and academic researchers.