We propose a new approach to handle the problem of portfolio optimization for non-life insurance company incorporating the solvency capital requirement (SCR), market views and their confident levels, several equality and inequality real-world constraints and transaction costs. We analyze two case studies: first, we consider a tri-objective optimization problem in which we minimize the Market SCR, the variance of the so-called basic own funds (BOF) and maximize the return of portfolio; secondly, we consider bi-objective optimization problem in which we minimize the variance of BOF and maximize the return of portfolio while considering the Market SCR as a constraint. We introduce a scenario-based framework in which the reference model is given by an internal model. By entropy pooling approach, we blended market views and their confident levels with the reference model to build the posterior distribution. The latter is used to compute the variance of BOF and the portfolio return. In both case studies, we obtain good results in term of risk-reward tradeoff and diversification.

我们提出了一种新方法来解决非人寿保险公司的投资组合优化问题，其中包括偿付能力资本要求（SCR），市场观点及其信心水平，现实世界中的几种平等和不平等约束以及交易成本。我们分析了两个案例研究：首先，我们考虑了一个三目标优化问题，在该问题中，我们最小化了市场SCR，所谓的基本自有资金（BOF）的方差并最大化了投资组合的回报；其次，我们考虑了双目标优化问题，在该问题中，我们将BOF的方差最小化并将投资组合的收益最大化，同时将市场SCR作为约束条件。我们介绍了一个基于场景的框架，其中参考模型由内部模型给出。通过熵池方法，我们将市场观点及其信心水平与参考模型进行了混合，以建立后验分布。后者用于计算BOF和投资组合收益的方差。在这两个案例研究中，我们在风险回报权衡和多元化方面均取得了良好的结果。