Portfolio Optimization under Fast and Slow Latent Mean-Reverting and Momentum Drift
We consider a class of partial information portfolio optimization problems in which a trader does not observe the true drift of a financial asset, driven by two latent stochastic factors operating at distinct time scales. We prove that under logarithmic, exponential, and power utility functions, the optimal strategies depend on the current price and on the difference between fast and slow exponentially weighted moving averages of prices, yielding a Moving Average Convergence Divergence (MACD) signal. The MACD structure is not imposed a priori: optimality is established over all admissible price-history-dependent strategies, and the divergence signal emerges endogenously as a parameter of the unconstrained optimal control.