Multi-Period Optimization¶
One-line definition¶
Choosing a portfolio (or trading path) to optimize wealth over multiple future periods, accounting for how today's trade affects tomorrow's opportunity set — as opposed to single-period optimization, which re-solves from scratch each period with no memory of the path.
Why PMs care¶
Single-period optimization implicitly assumes rebalancing is free and that today's decision doesn't affect what's achievable tomorrow — false whenever transaction costs, tax lots, or liquidity constraints carry across periods. A single-period-optimal trade today can be a bad trade once its effect on next period's costs and constraints is priced in.
Why this isn't implemented in src/pm¶
Genuine multi-period optimization is a stochastic control / dynamic programming problem — it needs a model of how the opportunity set evolves and a way to solve (or approximate) the resulting Bellman equation. That's a different category of tool from the closed-form and single-period-convex functions this repo builds elsewhere.