As network anchors and pundits appeared stunned at Donald Trump's Electoral College victory Tuesday night, it was obvious much political polling data had missed the mark.
Also most prediction markets—often touted as a better indicator for election outcomes—had widely predicted a Hillary Clinton victory, including one that had put her probability of victory at 83 percent.
"Prediction markets now have a lot of egg on their faces with the presidential election and the Brexit vote earlier this year," said Koleman Strumpf, a University of Kansas professor of business economics.
Prediction markets, or information markets, are exchange-traded markets that allow trading on the outcome of events, and the prices of the exchange can be an indicator of what a crowd thinks a probability of an outcome will be. Strumpf, who is associate editor of the Journal of Prediction Markets, said the markets were useful in elections when they cropped up about 100 years ago, especially in the absence of public polling.
After Tuesday's election, economists and data scientists will likely examine specifically what went wrong with most prediction markets and whether they have reached their limits at being useful, he said.
As far as the U.S. presidential election, it's possible polling data too heavily influenced decisions in the prediction markets.
"If apparently the information was bad, then the markets weren't able to figure that out," Strumpf said. "However, the markets' job is to see through that typically. For lack of a better expression, it seemed everything became an echo chamber. People got convinced of the inevitability of Clinton winning, and they were unwilling to fully weight a set of circumstances that could lead to what actually did happen, Trump winning the Electoral College majority."
Provided by University of Kansas