本文成功建構預測市場的最適價格門檻，作為判定預測事件是否發生的依據， 將事件發生的機率預測（probabilistic forecasting）轉換成事件發生與否的類別預測 （categorical forecasting），作為公共政策與企業決策的重要依據。以2006-2011 年的 未來事件交易所7,881 個預測事件樣本分析，本文透過計算所有價格的準確率尋找最 適價格門檻，並且比較價格50 門檻與最高價門檻準則的準確率。本文發現：全部預 測事件樣本的最適價門檻為價格60，其判定預測事件的準確率比價格50 門檻與最高 價門檻準則都要高，具有統計上的顯著性，並且通過樣本外測試。

We successfully construct a methodology to find optimal price thresholds for prediction markets as the basis of categorizing the occurrence of the prediction events. This method enables us to transform the probabilistic forecasting of event occurrence to the categorical forecasting, thereby creating an important reference for public policy and corporate decision-making. Based upon 7,881 samples of prediction events from the Exchange of Future Events between 2006 and 2011, we determine optimal price thresholds and compare the accuracy rates with those obtained by two benchmark rules. This paper finds that the price of 60 is the optimal price threshold for the all sample studied, which yields an accuracy rate higher than those obtained by the common threshold of 50 and the highest-price threshold. The superiority is statistically significant and is also robust in outof- sample tests.