Summary: | 碩士 === 東吳大學 === 經濟學系 === 99 === Kelly's investment approach was originally developed for betting in gambles. The payoff probability distribution of any gamble is fixed. Kelly Criterion gives a growth-optimal bet size (ratio) for gamblers. Vince (2007 and 2009)'s LSM model applies Kelly’s investment approach to financial assets. The payoff probability distribution of any financial asset will change, however, with the updating of price information. The estimated optimal bet ratio, in turn, changes accordingly.
Kelly's investment approach has several important features. First, for any gamble, there is a duality between the optimal bet ratios of the gambler and the banker. That is, the ratio of two players’ bet sizes associates, in certain inverse manner, with that of two players’ unit bet amounts. Second, the optimal bet size is inversely determined by the ratio of mean to standard deviation of the probability distribution. Third, for a given value of mean, the optimal bet ratio of fat-tailed distribution is lower than that of normal distribution.
The empirical findings show that investments in TW Index and S&P 500 Index, basing on the optimal ratios formed by LSM model, result in lower arithmetic and geometric mean returns than those earned by the buy-and-hold policy. The performances of the active strategies formed by LSM model are particularly significantly worse. The evidence suggests that Kelly’s investment approach applied to financial assets is like some kind of technical analysis. In well-functioning capital market, the performance of the active strategies formed by technical analyses is often outperformed by the passive buy-and-hold strategies.
|