Carbon-Penalised Portfolio Insurance Strategies in a Stochastic Factor Model with Partial Information

Battle Count: 134

The paper is primarily focused on portfolio insurance strategies for institutional investors (pension funds, insurance companies) rather than high-frequency or algorithmic trading. However, the stochastic control framework, factor model with partial information, and carbon-penalised allocation rules are relevant to systematic portfolio construction and risk-managed trading strategies. The dynamic multiplier and portfolio composition rules could inform quantitative allocation engines. The partial information framework (Kalman filtering of latent factors) is directly applicable to quantitative trading systems that estimate unobservable market states.

Implementation requires solving a system of coupled Riccati ODEs (for CRRA) or closed-form expressions (for log utility), implementing Kalman filtering for the Ornstein-Uhlenbeck factor, computing the Cholesky decomposition of the correlation matrix, and simulating the stochastic differential equations for the cushion process. The full-information case is analytically tractable, but the partial-information case requires numerical ODE integration and filtering. The multi-dimensional nature (n stocks + 1 factor) adds computational complexity. The admissibility conditions and existence proofs require careful parameter calibration.

The paper provides complete mathematical derivations, explicit parameter tables (Table 6.1), and detailed numerical experiments. All ODE systems and optimal control expressions are given in closed form. However, no code repository is provided, and the numerical implementation details (e.g., ODE solvers, simulation methods) are not fully specified. The sensitivity analysis in Appendix D provides additional robustness checks.