Traditional multi-armed bandit problems are geared towards finding the arm with the highest expected value -- an objective that is risk-neutral. In several practical applications, e.g., finance, a risk-sensitive objective is to control the worst-case losses and →