Using Prediction Markets to Forecast the Future
https://www.vox.com/future-perfect/2019/4/5/18290870/forecasting-tetlock-prediction-markets-betting
Today in class, we were introduced to prediction-markets. According to this Vox article, people can bet on almost anything with prediction markets such as if “migration from Central America will increase” or if “this new cancer study will fail to replicate” (Piper). If the event one wants to bet on does not exist in the prediction-market yet, they can introduce it into the prediction-market and bet on the event. Experts believe that prediction-markets have the potential to be used for accurately forecasting the future. Unfortunately, there are a couple of reasons why prediction-markets are not as mainstream as they have the potential to be. One of these reasons is that there are “legal and logistical hurdles” such as many countries “restricting gambling” or banning betting on “political and internal events” (Piper). Another reason is that many times, they are inaccurate. For example, betting markets had the probability that Great Britain would leave to be around 20% during the 2016 Brexit Referendum and a 35% chance that Donald Trump would win the presidency during the 2016 election. Considering that these two events are significant events that many people bet on, they should have been much more accurate in order to effectively forecast the future.
In section 22.3 of the textbook, Easley and Kleinberg alludes to the idea of the “wisdom of crowds” and how “the aggregate behavior of many people, each with limited information, can produce very accurate beliefs” (Easley, Kleinberg) If this idea is true, why were the forecasts for the 2016 US presidential election and Brexit Referendum so inaccurate? This can be answered by one of the limitations to the wisdom of crowds: “all beliefs are equally weighted” (Easley, Kleinberg). Kleinberg mentions the example of horse races and how state prices place more weight on wealthy betters compared to poorer betters because wealthier betters are probably wealthier because their bets are more accurate. The Vox article also supports this idea by mentioning a study (Jason Dana et al.) which found that surveying people and weighing their answers based on their “track record” were significantly more accurate than without weighing answers because “markets have a built-in mechanism to take people who are correct more often more seriously” (Piper). By conducting both polls with weighted answers and prediction markets, the study found that this combination predicted the most accurate estimates.