Sequential optimization of time-varying weights in Double Linear Policy (DLP)

Develop a sequential optimization methodology to determine time-varying weighting functions w_L(k) and w_S(k) in the Double Linear Policy trading framework, computed dynamically from available information at each time step rather than relying on prespecified weight schedules, so that the policy can adapt to time-varying market conditions.

Background

The Double Linear Policy (DLP) extends Simultaneous Long-Short controllers and preserves Robust Positive Expectation (RPE) under optimized constant weights or admissible prespecified time-varying weights. Prior work demonstrated survivability and RPE under time-varying weights but did not provide a principled way to generate them via optimization; instead, schedules were specified exogenously.

The explicit unresolved issue concerns selecting these weights dynamically in real time, as heuristic or backward-looking calibration approaches may not adapt to changing market conditions. The present work proposes an SMPC-based approach, highlighting that the broader problem of principled sequential optimization for DLP weights had remained open.

References

Despite these advances, the question of how to optimally select weights in a dynamic environment remains an open challenge.

— Dynamic Weight Optimization for Double Linear Policy: A Stochastic Model Predictive Control Approach  (2604.00415 - Hong et al., 1 Apr 2026) in Section 1, Introduction (p. 1)