Smith's (1962) demonstration that prices and allocations quickly converge to the competitive equilibrium in the continuous double auction (CDA) remains one of the most important results in experimental economics. Market experiments and exchange models have added considerably to our knowledge of how markets reach equilibrium, and how they respond to disruptions. Perhaps the best-known model of exchange in CDA market experiments is the random behavior 'zero-intelligence' (ZI) model by Gode and Sunder (1993). They argue that the CDA generates efficient allocations and 'convergence of transaction prices to the proximity of the theoretical equilibrium price,' provided only that agents meet their budget constraints. We demonstrate that prices do not converge in their simulations. Their budget constraint requires that a buyer's currency never exceeds her commodity value, which is an unnatural restriction. Their conclusion that market efficiency results from the structure of the CDA independent of traders' profit seeking behavior rests on their claim that the constraints that they impose are a part of the market institution. We show that actually they impose individual rationality. Misinterpretation of this behavioral constraint has led to unproductive debate on market adjustment processes.