Short-term power system operational planning problems that consider multi-stage uncertainties pose significant challenges, not only in the design of tractable optimization frameworks for implementing them, but also in the testing and benchmarking of such frameworks. This paper presents an implementation using the open-source MATPOWER Optimal Scheduling Tool (MOST) to study and compare a stochastic day-ahead, security-constrained unit commitment problem with a more traditional deterministic approach. The comparison is based on a testing methodology for day-ahead plans designed to produce expected performance estimates with minimal biases from modeling assumptions. Emphasis is given in the proposed stochastic approach to explicit modeling of the operational characteristics of the technologies available, their spatial and temporal coupling, and the regulatory constraints that assure reliability and adequacy. The problem formulations and testing methodology are described and simulation results from MOST are presented, with discussion of implications for future market design. All of the code and data to replicate the simulations is provided online.