Stochastic factors can matter: improving robust growth under ergodicity

Battle Count: 85

Highly relevant to quantitative trading. The paper directly addresses portfolio optimization under drift uncertainty, which is a central problem in quantitative finance. The pairs trading application is directly applicable to statistical arbitrage strategies. The framework provides actionable insights on when and how to incorporate stochastic factors into trading decisions, with explicit formulas for optimal strategies. The CTOU model robustification is directly relevant to practitioners in mean-reversion trading.

High complexity. Requires solving a collection of d-dimensional PDEs (one for each state y of the factor process). The general framework involves calculus of variations, generalized Dirichlet forms, and Fokker-Planck equations. However, in the Gaussian/OU case, explicit closed-form solutions are available. The dimension reduction (PDE is d-dimensional despite (d+m)-dimensional state space) helps computationally. Online solving per y-slice is feasible. Fat-tailed and stochastic volatility extensions require more complex numerical methods.

Code for all numerical results is freely available on GitHub. The paper provides explicit formulas for the Gaussian/OU case and detailed derivations. Assumptions are clearly stated and verified in examples. However, the general PDE solution requires numerical methods not fully specified in the paper.