Equity portfolios generated by functions of ranked market weights (Q5957681)

Functionally generated portfolios were introduced in \textit{R. Fernholz} [J. Math. Econ. 31, No. 3, 393-417 (1999; Zbl 0944.91025)] with the entropy-weighted portfolio and in more general form in \textit{R. Fernholz} [Portfolio generating functions. In: Avellaneda, M. (Ed.) Qualitative analysis in financial markets. River Edge, NJ: World Scientific (1999)]. It was shown that certain functions of the market weights in an equity market generate dynamic portfolios. Properties of functionally generated portfolios are different from the properties of usual portfolios considered in the literature. Functionally generated portfolios are not optimal in the sense of \textit{H. Markowitz} [J. Finance 7, 77--91 (1952)] or \textit{R. Merton} [Rev. Econ. Stat. 51, 247--257 (1969)] and they can not be expected to be asymptotically optimal in the sense of \textit{L. Breiman} [Proc. 4th Berkeley Symp. Math. Stat. Probab. 1, 65--78 (1961; Zbl 0109.36803)] or \textit{T. M. Cover} [Math. Finance 1, No. 1, 1--29 (1991; Zbl 0900.90052)]. Functionally generated portfolios are like the hedging portfolio corresponding to the Black-Scholes option pricing formula [see \textit{I. Karatzas} and \textit{S. Shreve}, Methods of mathematical finance. Applications of Mathematics. Berlin: Springer (1998; Zbl 0941.91032)]. In this paper functions of the ranked market weights are considered. It is shown that a positive twice continuously differentiable function of the ranked weights generates a dynamic equity portfolio. For such a portfolio the return relative to the market follows a stochastic differential equation that decomposes the relative return into two components: the logarithmic change in the value of the generating function, and a drift process that includes semimartingale local times. Two applications of the method are presented. The first one provides an explanation for the size effect, the observed tendency of small stocks to have higher long-term returns than large stocks. The second application provides a rigorous mathematical analysis of the behavior of diversity-weighted indexing, a new type of passive equity strategy that is currently being used for actual investment.