Academics explain how they trumped the pollsters
Source: Taipei Times
Reporter: Max Hirsch
APPLIANCE OF SCIENCE: Taiwan’s first prediction market tapped into the “wisdom of the masses" to forecast the results of recent mayoral polls
While media polls failed miserably in predicting the results of the Taipei and Kaohsiung mayoral elections, one group of academics and political soothsayers prophesied the outcome with stunning accuracy.
Their nearly dead-on predictions were no accident, they insist, nor were there any crystal balls involved.
Just good science was behind their forecasts, they said, adding that they would soon unveil more on-the-mark political predictions — including the winner of next year’s presidential election.
“It’s like a futures exchange market," said Academia Sinica political analyst Lin Jih-wen (林繼文), referring to the Center for Prediction Markets, run jointly by National Chengchi University (NCCU) and Swarchy Co, at a press conference touting the center yesterday.
“The same market mechanisms as in a bourse are harnessed to integrate biases and diverse outlooks on the future and produce a specific prediction," he said.
Like prediction markets in the US, which boast steady track records of accurately predicting everything from the next president to the next Oscar winner, Taiwan’s first prediction market is a speculative bourse whose assets’ final cash value is tied to a specific future event.
Speculators bet on various possible outcomes whose market value is interpreted as the probability of that outcome occurring, said Swarchy Co official Liu Chia-kai (劉嘉凱).
The final market projection for Democratic Progressive Party (DPP) Taipei mayoral candidate Frank Hsieh’s (謝長廷) share of the vote in December’s election, for example, was off by merely a few percentage points, said panelists.
The market’s predicted winner of the Taipei mayoral election — Chinese Nationalist Party (KMT) candidate Hau Long-bin (郝龍斌) — was similarly pegged by speculators as earning a vote share that missed the mark by only one percentage point.
“Contrast that with media polls," Liu said. “They were all off in their forecasts of voting shares by around ten percentage points, and even failed to predict the winner."
So why was the prediction market so accurate?
Center director and NCCU economics professor Lin Hsin-yi (林馨怡) told reporters that when a large group of informed individuals with a material stake in how the event that their betting on, unfolds, they tend to actively and squarely assess the circumstances.
Liu called such statistical accuracy “tapping the wisdom of the masses."
Polls, on the other hand, rely on the responses of often apathetic or even untruthful participants, said Academia Sinica researcher Hsu Yung-ming (徐永明), adding that media should analyze both polls and prediction market trends in sizing up political developments.
“Polls are useful because they can roughly tell us the political thinking among certain segments of society, while prediction markets can tell us, with a reasonable degree of accuracy, what’s going to happen," Lin Jih-wen said, adding that prediction markets’ weakness lies in their inability to demonstrate why they have reached certain conclusions.