The investment of insurance funds is one of the main things for the insurance. The effective investment of insurance funds has very important effect on its operation and development. In recent years, studying the optimal portfolio of insurance funds has become a hot issue. We know that any investment has its risky. So how to choose an effective and low risk portfolio is particularly important. The research methods of the insurance portfolio mainly are mean variance theory of Markowitz, stochastic control theory and Var method. For these methods, there are three main criteria:The first is Markowitz's mean variance criteria; The second is to maximize the effectiveness of investor; The third is to minimize the probability of bankruptcy.The difference of the classical statistics and the Bayesian statistics is:the classical statistics infer based on model information and sample information. Besides the two information, the Bayesian theory also uses unknown parameter prior information.In this paper, we first introduce three ways of the optimal portfolio selection based on the theory of mean variance, The first is to minimize portfolio's risk, for a given minimal (required) return; The second is to maximize portfolio's expected return, for a given maximal risk; The third is to maximize investor's expected utility function. The second, we discuss minimizing portfolio's risk of insurance funds for a given minimal (required) return, frame the corresponding model and estimate the parameters of the model with the Maximum likelihood method. The emphasis of the paper is the third method, we first construct the utility function of insurance portfolio. The next we solved the investment weight in the condition of maximizing investor's expected utility function. And we estimate the unknown parameters with Bayesian theory in the two different priors(diffuse prior and proper prior). Last, we select the year yield of index in the database of Ruisi, and comparing these methods according to the efficient frontier criteria of mean and covariance. |