u(W) = –cW + Wb– = (2cE(W) +1). Moreover, optimizing the expected value of the

In particular, Table II reports the threshold values c1 and c2 respectively for the Gaussian and the stable distributed index returns. Generally, c1 and c2 are different. Thus, even when we consider the utility functional (6), the investors' choices in the Gaussian case could be completely different from the stable one. The three indexes analyzed present the threshold risk aversion parameter c1 always lower than c2. Therefore, when c (c1, c2) every investor who maximizes the utility functional (6): a) chooses the riskless asset, if the data fit the normal distribution; b) chooses the risky asset, if the data fit the stable distribution. Outside the interval (c1, c2) the choices do not differ under different distributional assumptions (Stable/Gaussian) and every investor who maximizes the utility functional E(W) – cE (|W – E(W)|) c) chooses the risky asset, if c < c1; d) chooses the riskless asset, if c > c2.