Via Joseph Sterlynne on the Extropy list: NEC Research Institute Technical Report #2000-168. A brief version appears in Science 291: 987-988, February 9 2001 (Letters). The Power of Play: Efficiency and Forecast Accuracy in Web Market Games David M. Pennock Steve Lawrence C. Lee Giles Finn Årup Nielsen1 http://artificialmarkets.com/am/pennock-neci-tr-2000-168/ Abstract We analyze the efficiency and forecast accuracy of two market games on the World Wide Web: the Hollywood Stock Exchange (HSX) and the Foresight Exchange (FX). We quantify the degree of arbitrage available on HSX, and compare with a real-money market of a similar nature. We show that prices of HSX movie stocks provide good forecasts of actual box office returns, and that prices of HSX securities in Oscar, Emmy, and Grammy award outcomes constitute accurate assessments of the actual likelihoods that nominees will win. Similar investigations reveal that FX securities prices serve as reliable indicators of uncertain future events. We argue that, in certain circumstances, market simulations can furnish some of the same societal benefits as real markets, and can serve as acceptable substitute testbeds for conducting experiments that would otherwise be difficult or impossible. Keywords: analysis of artificial markets, World Wide Web market games, market simulations, forecast accuracy, economic efficiency, arbitrage, Hollywood Stock Exchange, Foresight Exchange, utility for intangibles Introduction The core service of a market is to facilitate the exchange of items between individuals. The use of prices for these items, denominated in a common currency (e.g., US dollars), simplifies trading across multiple markets, alleviating the combinatorial nature of direct barter. Prices reflect an agreement between buyers and sellers, and serve as a quantitative measure of the value of the item being exchanged, as compared to other marketable items. When markets attract broad participation, prices can encode the sum total of a large amount of disparate and distributed information. The prices reflect, in a very real sense, the consensus opinion of a myriad of informed and well-motivated traders. As such, even nonparticipating observers may stand to benefit from the informational value of market signals. As an example, the odds in a horse race, determined solely by market forces at the track, can be viewed as assessments of the likelihoods that the various horses will win. Empirical studies verify that odds on horses do indeed match very closely with their observed frequencies of winning [1,18,19,20,22]. As traditional markets expand onto electronic platforms, and as new electronic marketplaces emerge, price information will be available and accessible in quantities previously unimaginable. Nevertheless, markets will still only cover a miniscule fraction of arenas for which informed forecasts might be valuable or interesting. Many barriers exist for the establishment of new markets, including high costs, government regulation, and the threat of lawsuits. Artificial markets, on the other hand, suffer from no such difficulties. Web market games, in particular, often feature moderate operating costs for setup, maintenance, advertising, searching, and transacting, and benefit from worldwide audience potential. Permission is not required from government authorities or regulatory officials. Lawsuits are much less of a concern. There is little need for carefully crafted disclaimers or facilities for dispute resolution. Users can remain anonymous, and record keeping can be somewhat lax. All of these factors have contributed to a growing prevalence of market games on the web, some enjoying widespread popularity. Of course, artificial markets cannot satisfy societal demand for the exchange of items. However, in this paper we present evidence that some market simulations can function reasonably well in the dual role as aggregators and disseminators of information. Theories of market equilibrium, including the rational expectations theory of information propagation, usually depend on the assumption that participants maximize expected utility, where utility is derived from consumables or monetary equivalents. Indeed, laboratory economics experiments in which subjects are not ``paid to play'' are often questioned on the grounds of a lack of true incentives. In a game without monetary backing, utility is presumably extracted solely from entertainment value, educational value, bragging rights, and/or other intangible sources. Does market efficiency simply break down under these conditions, or can non-monetary rewards actually drive price coherence, information aggregation, and forecast accuracy? We find evidence that, in some cases, they can and they do. In Section 3 we quantify price coherence on the Hollywood Stock Exchange (HSX). Equivalent portfolios trade at reasonably consistent prices, and, over time, large inefficiencies disappear, as players presumably take advantage. In Section 4 we evaluate the collective competence of traders on HSX and on the Foresight Exchange (FX), by measuring the prescient value of market prices. In doing so, we find that HSX stock prices are reliable indicators of what movies will do well at the box office, that HSX award option prices provide accurate assessments of which nominees will likely win entertainment awards, and that FX prices constitute accurate probabilistic judgments for a variety of uncertain future events. -- Use e-gold? Send me two cents: http://2cw.org/257121&[EMAIL PROTECTED] Read the _Wall Street Performer Protocol_: http://www.openknowledge.org/writing/open-source/scb/