Sequential Audit Sampling for Finite Populations with Exact and Simulation-based Guarantee

Battle Count: 64

This paper is focused on financial statement auditing and control testing, not on trading strategies, market microstructure, or quantitative finance in the trading sense. While it falls under q-fin.ST, its application domain is audit methodology rather than investment or trading decisions. The sequential testing framework has theoretical parallels to sequential decision problems in trading, but the paper does not address any trading application.

The exact hypergeometric recursion (Algorithm 1) is straightforward to implement for one-dimensional state spaces with O(n^2) operations. However, the boundary construction logic, handling of decision times, monotonicity verification, and the Monte Carlo extension with Clopper-Pearson bounds add moderate complexity. The paper provides sufficient algorithmic detail for implementation but requires careful handling of edge cases (adjacent boundaries, infeasible regions).

The paper provides Algorithm 1 for exact boundary computation, detailed numerical examples with specific parameter settings (n=100, r=0.20, theta_H=0.05, alpha=beta=0.05), and references to supplementary material with additional calculations. The CMS CERT data is publicly available. However, no explicit code repository is linked in the main text.