RSOME案例：多阶段投资计划的随机规划实现

RSOME: Robust Stochastic Optimization Made Easy

---

Multi-Stage Stochastic Financial Planning

In this example we consider the multi-stage financial planning problem discussed in page 23 of Birge and Francois (2011). As a multi-stage model problem, the here-and-now decision allocates the total wealth $d$ between two investment types: stocks (S) and bonds (B). Each investment type may have a high return and a low return as two possible outcomes. It is assumed that the high returns for stocks and bonds are 1.25 and 1.14, and the low returns are 1.06 and 1.12, respectively. These outcomes are independent and with equal likelihood, so throughout the subsequent stages, we would have eight scenarios with equal probabilities. The random return outcomes of stocks and bonds are represented by a scenario tree shown below.

Besides the deterministic equivalent of the stochastic model, the financial planning problem can also be formulated as a distributionally robust optimization problem where the decision tree is represented by the following ambiguity set,

$$\mathcal{F}=\left\{\mathbb{P}\in {\mathcal{P}}_0({\mathbb{R}}^{3\times 2}\times [S])|\begin{array}{llcc}& (\widetilde{z\text{z}z},\tilde{s})\in \mathbb{P}& & \\ & \mathbb{P}[\widetilde{z\text{z}z}\in {\mathcal{Z}}_{\tilde{s}}|\tilde{s}=s]=1,& & \forall s\in [S]\\ & \mathbb{P}[\tilde{s}=s]=1/S& & \end{array}\right\},$$

with the scenario number $S=8$ and the support $\mathcal{Z}_s=\left{\hat{\pmb{z}}_s \right} $. The financial planning problem is thus written as

max ⁡