The decision of the selection strategy of venture project portfolio is an vital concern for venture capital firms. Venture capital firms make full use of the limited resources to create value by selecting and executing venture projects portfolio, and the behavior preference of venture investor directly affects the optimal outcome of venture project portfolio. Usually, the value of venture project is uncertain, so venture investor must make investment decision based on prior estimation of future value of venture projects. Under the framework of prospect theory, this paper constructs a portfolio optimization model of venture projects considering the psychological characteristics of venture investors, and proposes a Bayesian method to deal with the uncertainty of value estimation in venture project portfolio, and adopts Monte Carlo method to simulate the model as a linear integer programming problem. This study finds that, compared with venture project portfolio directly based on ex ante value estimation, Bayesian modeling of the uncertainty of the project's value prior estimation can give a more accurate value estimate, and utlize the obtained revised estimates to make portfolio decisions, that can help to select a project portfolio with a higher expected utility, eliminate the expected interval between the expected utility of beforehand and the expected utility of post-implementation, and reduce the degree of disappointment of venture investor's decision-making.
