Researchers develop smarter way to pick the best option

A new algorithm developed by researchers could transform how organizations make high‑stakes decisions, helping computers pinpoint the best option faster and with greater accuracy.

Ye Chen, an assistant professor of applied statistics and operations research for the Allen W. and Carol M. Schmidthorst College of Business at Bowling Green State University, and co‑author Ilya O. Ryzhov created an algorithm called BOLD, short for Bayesian Optimal Large Deviations. The method addresses a widespread challenge in business and science: when there are many potential choices, how do decision‑makers determine which ones are worth testing? BOLD offers a smarter, more strategic way to identify the most promising options.

The method applies to scenarios such as evaluating factory layouts, inventory policies or plant breeding traits. For example, a manufacturer comparing several production line designs could use BOLD to quickly pinpoint the setup most likely to improve efficiency, reducing the time and cost typically associated with trial‑and‑error testing. 

The algorithm uses probability theory to balance two essential goals: exploring new possibilities and concentrating on those that show the most potential. This balance helps organizations identify the best choice with high confidence, all while conducting fewer tests.

The research is especially significant as more industries rely on computer simulations to guide decision‑making. More efficient testing methods lead to faster, stronger outcomes. Manufacturing companies, supply chain managers and agricultural researchers are among those who could benefit from this approach.

Chen received the INFORMS Outstanding Simulation Publication Award for this work, the highest honor given for a single research publication in the simulation field. Sponsored by the INFORMS Simulation Society, the award was presented at the Winter Simulation Conference, held Dec. 7‑10, 2025, in Seattle.

Balancing Optimal Large Deviations in Sequential Selection is avaiablae via Informs.