How is the structure of trade flows for a given good organized? Is it a dense, widespread network with many links, or is it a centered network, organized around a hub that centrally coordinates the flows? Does it have a regional structure or a world-wide coverage? These questions might have very different answers according to the type of goods that are traded. The existing trade models suggest that for standardized goods potential partners can be many, but eventually only one (the one offering the best price) should be selected, therefore relatively few (unidirectional) trade links will appear between countries. Instead, for goods produced in many varieties and qualities, there will be many trading partners, with many multiple links. For goods with low transport costs, partners can be either far away or close, while for goods with high transport cost, geographic proximity of partners will be important, giving rise to regional clusters. In high-tech, sophisticated productions, the technological proximity of partners can give rise to selected trade flows around a technologically-advanced center. The structure of the trade network in terms of density, centralization and clustering can have a direct impact on the competition between countries and on the formation of the international price for a given good, by defining the size of the market in terms of potential demand and supply. In this paper we test whether different goods give rise to trade networks with different structural characteristics, as a simple theoretical framework of trade relations would suggest. Network analysis indices are used to identify the different characteristics of trade networks in terms of density, clustering and centrality, and we assess whether these networks are regional or world-wide in scope. The results of this analysis are then used to define the market characteristics for different types of goods.