Theoretical challenges

Some academic research has focused on potential flaws with the prediction market concept. In particular, Dr. Charles F. Manksi of the Northwestern University Department of Economics published a paper in 2004, “Interpreting the Predictions of Prediction Markets”, [2] in which he attempts to show mathematically that under a wide range of assumptions the “predictions” of such markets do not closely correspond to the actual probability beliefs of the market participants unless the market probability is near either 0 or 1. Manski suggests that directly asking a group of participants to estimate probabilities may lead to better results. However, Steven Gjerstad (Purdue) in his paper “Risk Aversion, Beliefs, and Prediction Market Equilibrium” [3] has shown that prediction market prices are typically very close to the mean belief of market participants if the distribution of beliefs is smooth (as with a normal distribution, for example). Justin Wolfers (Wharton) and Eric Zitzewitz (Stanford) have obtained similar results, and also include some analysis of prediction market data, in their paper “Interpreting Prediction Market Prices as Probabilities” [4]. In practice, the prices of binary prediction markets have proven to be closely related to actual frequencies of event in the real world. Relevant data has been published in Pennock et al’s “The real power of artificial markets” [5] (Science, 2001) and Servan-Schreiber et al’s “Prediction Markets: Does Money Matter?” [6] (Electronic Markets, 2004).

Prediction markets also suffer from the same types of inaccuracy as other kinds of market, i.e. liquidity or other factors not intended to be measured are taken into account as risk factors by the market participants, distorting the market probabilities. There can also be direct attempts to manipulate such markets. In the Tradesports 2004 presidential markets there was an apparent manipulation effort (an anonymous trader sold short so many Bush 2004 presidential futures contracts that the price was driven to zero, implying a zero percent chance that Bush would win. The only rational purpose of such a trade would be an attempt to manipulate the market in a strategy called a “bear raid”. The manipulation effort failed, however, as the price of the contract rebounded rapidly to its previous level.) As more press attention is paid to prediction markets, it is likely that more groups will be motivated to manipulate them. However, in practice, such attempts at manipulation have always proven to be very short lived. In their forthcoming paper entitled “Information Aggregation and Manipulation in an Experimental Market” (2005) [7], Hanson, Oprea and Porter (George Mason U), show how attempts at market manipulation in fact end up increasing the accuracy of the market because they provide that much more profit incentive to bet against the manipulator.

Prediction markets may also be subject to speculative bubbles. For example in the year 2000 IEM presidential futures markets a flood of new traders in the final week of the election caused the market to gyrate wildly, making its “predictions” useless.

A common belief among economists and the financial community in general is that prediction markets based on play money cannot possibly generate credible predictions. However, the data collected so far disagrees. Pennock et al (Science, 2001) analyzed data from the Hollywood Stock Exchange and the Foresight Exchange and concluded that market prices predicted actual outcomes and/or outcome frequencies in the real world. Servan-Schreiber et al (Electronic Markets, 2004) compared an entire season’s worth of NFL predictions from NewsFutures’ play-money exchange to those of Tradesports, an equivalent real-money exchange based in Ireland. Both exchanges performed equally well. In this case, using real money did not lead to better predictions.

Some experimental systems are underway to provide data on alternatives to prediction markets that seek to avoid some of the theoretical pitfalls mentioned earlier. For example, polling firm TIPP Online has experimented with “national zeitgeist” questions which ask participants who they think will win rather than who they will vote for personally. This proved to be a more stable and accurate predictor in the 2004 US presidential race than traditional polls. Another experimental system is Owise which directly asks participants to estimate probabilities on a wide range of future events, and rewards accurate performance with status, titles, and small cash prizes. Owise functions as a hive mind or a kind of neural network in which each “neuron” is a human being whose predictions are assigned a weight based on past performance. In fact, this is not so different from what naturally happens in a prediction market where those who make good predictions do profit at the expense of those who make bad predictions, thus progressively increasing their relative influence on the market through how much money they can bring to bear to back up their predictions. There is currently not enough data and history to check how these alternatives will compare to prediction markets in terms of forecasting ability.

Commercial interest

• Hewlett-Packard pioneered applications in sales forecasting and now uses prediction markets in several business units. Mentioned in academic publications from HP Labs. Also mentioned in Newsweek [8] (October 2004)
• Corning, Eli Lilly, Abbott Labs, Siemens, Masterfoods, Arcelor and other global companies are listed [9] as NewsFutures customers.
• Intel mentioned in Harvard Business Review (April 2003) in relation to managing manufacturing capacity.
• Microsoft is piloting prediction markets internally.
• France Telecom’s Project Destiny has been in use since mid-2004, with very successful predictive behaviour.
• Google has confirmed that it uses a predictive market internally in its official blog [10].

This guide is licensed under the GNU Free Documentation License. It uses material from the Wikipedia.

This entry was posted on Monday, March 9th, 2009 at 11:58 am and is filed under Prediction market. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.