An evolutionary trade network game (TNG) is proposed for studying the interplay between evolutionary game dynamics and preferential partner selection in various market contexts with distributed adaptive agents. The modular form of the TNG facilitates experimentation with alternative specifications for trade partner matching, trading, expectation up dating, and trade strategy evolution. Experimental results obtained using a C++ implementation suggest that the conventional optimality properties used to evaluate agent matching mechanisms in static market contexts maybe inadequate measures of optimality from an evolutionary perspective.