A Numerical Approach to Profile Investor Preferences from Option Prices

Abstract

This paper introduces a behavioural model and an algorithm that allow define classes of investors and draw the size each of them from financial data. The nonparametric pricing kernel estimated from stocks and options quotes allows to derive an estimate of the market utility. At the micro level it is assumed that each individual perceives lower future returns and higher future returns differently, and at a given threshold the individual switches from one attitude to the other. This switching point is peculiar of each class. The aggregate of the individual utilities must have the same features of the estimated market utility function, and two random search algorithms to compute the optimal aggregation are proposed and compared. Both computation techniques provide a similar distribution of investor classes. When the markets are bearish, even negative future states are perceived as high. On the other hand when markets are performing well, the switch to the high perception occurs only for bigger returns. For stock markets without a clear trend there is no predominance of a single class, investors are split between ’"early"’ and ’"late"’ switchers.