Discrimination-Insensitive Pricing

Battle Count: 61

This paper is primarily focused on insurance and credit pricing fairness rather than quantitative trading. While distortion risk measures (VaR, ES) are used in both contexts, the core contribution—discrimination-insensitive pricing via KL divergence optimization—is specific to actuarial and credit risk applications. The mathematical tools (KL barycentre, sensitivity analysis) could theoretically be adapted to trading contexts, but the paper does not address trading strategies, portfolio construction, or market microstructure.

The theoretical framework involves solving constrained distributional optimization problems with Lagrange multipliers, requiring root-finding for systems of equations derived from cumulant generating functions. The barycentre reconciliation adds another layer of optimization. The numerical implementation requires Monte Carlo simulation (1M+ samples), iterative root-finding with convergence criteria, and handling of discrete/categorical covariates via mollification. However, the closed-form representations of RN derivatives simplify the actual computation once Lagrange multipliers are found.

The paper provides detailed mathematical formulations, closed-form solutions, and a numerical case study with specified parameters (500,000 simulated samples, specific distributional assumptions, ES tolerance α=0.9, risk-loading c=0.2). However, no code repository is provided, and the numerical implementation relies on root-finding procedures with Monte Carlo simulation (1,000,000 realisations, 50 accepted samples). Reproduction would require implementing the Lagrange multiplier optimization and barycentre reconciliation from scratch.