A stochastic SIR model for cyber contagion: application to granular growth of firms and to insurance portfolio

Battle Count: 78

The paper is primarily focused on cyber insurance and risk management rather than trading strategies. However, it has indirect relevance to quantitative trading through: (1) systemic risk assessment that could affect market volatility, (2) firm-level revenue disruption modeling relevant to equity trading, (3) jump-diffusion processes and Cox processes that are also used in trading models, (4) aggregate loss estimation relevant to portfolio risk management. The granular model of firm growth and heavy-tail distributions are relevant to factor investing and risk factor modeling. The paper is more applicable to insurance quant and risk management than to algorithmic trading.

The model is highly complex with multiple interacting stochastic components: (1) multi-group SIR with 2K+K² stochastic CIR parameters (672 coefficients for K=12), (2) granular firm growth with correlated Brownian motions, (3) hybrid Cox-Bernoulli arrival process, (4) jump-diffusion revenue dynamics, (5) Monte Carlo simulation of AEP with Poisson event counts. Calibration requires forward simulation optimization. The mathematical framework involves stochastic ODEs, CIR processes, Cox processes, and multi-dimensional integration. Implementation requires careful handling of multiple filtrations (F^B, F^W, G) and conditional expectations.

The paper provides detailed mathematical formulations, calibration procedures (Algorithm 1, 2, 3), and parameter tables. However, no code repository is mentioned. The dataset (Pappers, GuidePoint Security reports) is publicly available but requires manual processing. The calibration procedure is well-documented but complex with multiple stochastic components. The LockBit ransomware data from May-July 2024 is publicly reported. Reproduction would require significant implementation effort given the multi-layered stochastic model.