An Empirical Investigation of the Option-Adjusted Realized Return

Abstract

The issue of meaningful evaluation of professionally managed portfolios remains to be resolved satisfactorily within the investment community. The fact that many of the current procedures for evaluating portfolio performance are deeply rooted in conventional mean-variance (M-V) analysis raises serious concerns from a theoretical perspective. The primary objective of this paper is an empirical investigation (as differentiated from a thorough empirical test) of an ordinal portfolio performance measure, called the Option-adjusted Realized (average rate of) Return or ORR, developed recently by Smith and Kokoska (1998). The ORR is a leverage- and risk-adjusted average realized rate of return that can be used directly in evaluating portfolio performance.

Using returns data for June 1992 to May 1998, we estimate ORRs for two “portfolios”—the CREF Stock Fund and a hypothetical “market index portfolio” whose composition is identical to that of the S&P 500 Index. Also, we estimate Sharpe and Treynor ratio values for each portfolio and compare rankings provided by these methods for the two portfolios with rankings provided by the ORR method. For the interval of time from June 1995 to May 1998, the rankings provided by the three methods are not consistent. The ORR rankings for this time period indicate the CREF Fund underperformed the S&P Index on a risk-adjusted basis. Additional partitioning of the data creates other multiple intervals or holding periods for which the evaluation results (ex post) support at least moderate likelihood of unambiguous inconsistency ex ante. We argue that, given our set of assumptions, the ORR rankings, founded in option-pricing theory, are more reliable than the others that are M-V based.