On Robust Optimal Linear Feedback Stock Trading (2202.02300v1)

Abstract: The take-off point for this paper is the Simultaneous Long-Short (SLS) control class, which is known to guarantee the so-called robust positive expectation (RPE) property. That is, the expected cumulative trading gain-loss function is guaranteed to be positive for a broad class of stock price processes. This fact attracts many new extensions and ramifications to the SLS theory. However, it is arguable that "systematic" way to select an optimal decision variable that is robust in the RPE sense is still unresolved. To this end, we propose a modified SLS control structure, which we call the {double linear feedback control scheme}, that allows us to solve the issue above for stock price processes involving independent returns. In this paper, we go beyond the existing literature by not only deriving explicit expressions for the expected value and variance of cumulative gain-loss function but also proving various theoretical results regarding {robust positive expected growth} and {monotonicity}. Subsequently, we propose a new {robust optimal gain selection problem} that seeks a solution maximizing the expected trading gain-loss subject to the prespecified standard deviation {and} RPE constraints. Under some mild conditions, we show that the optimal solution exists and is unique. Moreover, a simple graphical approach that allows one to systematically determine the optimal solution is also proposed. Finally, some numerical and empirical studies using historical price data are also provided to support our theory.