The feasibility of applying the "bootstrapping" methodology to the securities analyst's task of estimating future returns of common stocks is evaluated. Five analysts each estimated the returns of 35 securities on the basis of 22 information cues, selected independently by nonparticipating analysts. Stepwise regression was used to estimate bootstrapping models and actuarial models for all analysts. Using a subset of IS cross-validation securities, bootstrapping and actuarial models generally were superior to the analysts. The performance of a "composite" analyst was superior to the performances of the five participants. Results indicate the desirability of incorporating linear models into the analyst's judgmental task.In recent years two distinct but related approaches to the study of human judgment have been used. The first approach emphasizes the discovery of explicit descriptions of the form of the decision maker's judgment process. The objective of such research is to lay out in precise terminology (usually mathematical) the decision maker's process of integrating the available information into a conclusion. These efforts have resulted in complex simulation models (