A simulation and test of OptiMark's electronic matching algorithm and its simple variations for institutional block trading

OptiMark, an automated crossing system that matches institutional buyers and sellers of company stock, is simulated via a multi-agent system. The results are contrasted with two algorithm variations, which are based on a cruder rating system. Performance measures include the volume of stocks traded, the extent of price discrimination, trader utility, efficiency of matches, and processing time. Our results indicate that the cruder rating schemes perform as well as the original version for moderately sized markets. Yet, a test with human subjects implies that our crudest preference elicitation scheme appears simpler to use than the original scheme. We hope that this finding will help resolve some recent problems that OptiMark's system has faced concerning the difficulty of use.