In many practical investment situations the amount of available memory on stock data is extremely huge. Thus many investors are attracted to base their decisions on the information "currently available in their minds" (see [1, 2]). In the present paper various risk measurement models having application in the investment management are discussed. First we explain the concept of mean variance efficient frontier and Markowitz’s model to find all efficient portfolios that maximize the expected returns and minimize the risk. Markovian risk measures are also mentioned. Some measures of portfolio analysis based on entropy mean-variance frontier are studied. Risk aversion index and Pareto-optimal sharing of risk are explained. In view of these facts it is very interesting to study how the investor should make investments so that his total expected return is maximized and the risk of losing his capital is minimized. A maximum entropy model in risk sharing is proposed and applied to some problems.