This Master Thesis proposes an analytical model of commodity trading in the presence of market power, which is new to the literature. We establish the rational behavior of traders between two markets and derive price spread dynamics based on the trading volumes. We consider two commodity markets that quote different prices for the same product, and a trader capable of purchasing in one market to resell in the other, within a trading capacity. For simplicity, we assume that the inventory present between the two markets is constant, i.e., the amount stored is fixed (and is directly linked to the trading capacity). We first establish the structure of the optimal trading policy. This policy determines, given a realization of the current price spread between the two markets, how much (if any) volume must be bought/sold from/at the first market and sold/bought at/from the second market, so that the long-run average profit is maximized. When optimizing, the trader takes into account its impact on future spreads. We then describe the parameters of the optimal policy and we apply our model to kerosene prices in New York and Los Angeles, and illustrate the insights derived.