Maximum entropy model for forming an investment portfolio

The article considers the problem of allocating investments across a set of assets while applying lower requirements to the input data compared to common approaches. The need for researching this problem springs from limited abilities for acquiring information regarding possible outcomes of investment decisions in practice. The article assumes the uncertainty regarding the outcomes is premised on the probabilistic nature of the target value. At the same time, the only accessible information includes ranges of possible target values for each considered asset. The problem is framed within the concept of expected utility and is solved for the preferences of investors with constant risk aversion while assuming the considered assets are statistically independent of each other. The result of the problem solution is represented by the analytical expressions of the preferential allocation of the investments across the assets. The resulting expressions can be used directly for forming portfolios in making investment decisions. An illustration of using the proposed model is given. The generalized conclusions are formulated regarding the commonality and usability of the received results.