In this paper a Heterogeneous Agent Model for the cacao market price is extended and analyzed. It explicitly incorporates the recent financialization of commodity markets. It takes into account non-linear speculator behavior as an explanatory factor. The speculators are assumed to follow either a mean-reversion strategy or a trend following strategy and are allowed to switch strategies based on past performance. The fundamental price that guides the behavior of the speculator following a mean-reversion strategy is explicitly modeled with fundamental factors. The model is estimated on daily frequency data to provide detailed insight into price developments. The model estimates stay close to their theoretical expectations, though are not all significant. Forecast performance is similar to benchmark models amongst which a modified model that uses a moving average as fundamental price. The results further show a decrease of the fraction of speculators following a mean-reversion strategy after periods of overpricing. Based on these results, the conclusion is drawn that the cacao market price model evaluated here reveals significant dynamic speculator behavior. The behavior is difficult to reconcile with expectations raised by the literature. Further, it is concluded that HAM’s can be modeled to lay closer to economic reality by using an explicitly modeled fundamental price. Further improvements along the lines directed by the estimation results, might lead to increased statistical performance as well as modeled price dynamics more in line with expectations.