A Conceptual Framework for Studying Self-learning Agents in Recommerce Markets

Rainer Schlosser, Alexander Kastius

Abstract: In many markets, customers as well as retailers look for increased sustainability. Recommerce markets—which offer the opportunity to trade in and resell used products—are constantly growing and help to use resources more efficiently. To additionally manage the trade in and resell prices for used product versions is challenging for retailers as substitution and cannibalization effects have to be taken into account. An unknown customer behaviour as well as competition with other merchants regarding both sales and buying back resources further increases the problem’s complexity. Data-driven pricing agents offer the potential to find well-performing strategies and satisfy the need for automated decision support, particularly in online markets. As the training of such agents is typically data hungry and too costly to be performed in practice, synthetic test environments are needed to try out and evaluate self-learning pricing agents in different market scenarios. In this paper, we propose a conceptual approach for such a recommerce market simulation framework and its basic components. Further, we discuss requirements and opportunities to study self-learning strategies in synthetic markets.