Using Reinforcement Learning To Value Multi-Factor Option Mining Projects

In this study, we consider both one-factor and two-factor real options mining valuation problems. In the one-factor case, we further explore the use of RL to solve the build / abaondon problem we presented in \citeN{Lawryshyn23}. In the two-factor model we introduce a second process where the quantity (or quality) of the mined mineral is uncertain and, as mining proceeds, more is learned about the quantity available allowing for staged investment. We attempted to solve the two-factor "learn-as-you-go" (LAYG) problem using the EBF method previously with partial success. Our objective is to explore the opportunity to use RL for valuing realistic, computationally difficult real options problems. We note that this study is a work in progress.