本文以自信衡量指標—Brier 分數，衡量證券商對股價指數的預測是否過 度自信，並觀察過度自信與短期交易量的關係。從指數漲、跌與盤整三種走 勢觀察，發現：樣本券商對指數的預測過度自信，其程度呈週期起伏，以指 數波段中段的程度最高，故此期間指數最難預測﹔券商雖過度自信，但其自 營部未必增加交易量：對看漲過度自信時反而少買，看跌時卻多買少賣，預 期盤整時則傾向出售。這些交易方向與預期不符，故文獻中「過度自信時交 易量會增加」的說法失之簡化，應分成漲、跌與盤整三種預測來觀察。此結 果可能是因券商具專業和資訊優勢，使其可較快調整交易，故投資人如欲參考券商的資訊，當發現走勢與其預測不符，即宜調整交易策略，以免與自營 商對作，增加投資風險。

This paper investigates the forecasts made by securities firms for stock index in rising, falling, and oscillating situations separately and integrally to directly measure overconfidence with the Brier score, which is widely utilized in psychology but first employed in behavioral finance. The purposes of this paper are to examine: 1) Whether securities firms are overconfident and, if so, the characteristics of overconfidence; 2) the relationship between overconfidence and the short-term trading volumes of their dealer departments. Our findings are as follows：1) Overall, the sample firms are overconfident in forecasting index short-term trends; 2) overconfidence shows a cyclically undulating phenomenon while the medium sessions of rising and falling periods exhibit the highest overconfidence for firms and are, therefore, the most difficult ones to predict; 3) overconfident firms do not raise their trading volumes; instead, they buy less while forecasting a rising index, buy more while anticipating a declining index, and sell more while predicting oscillation. These results are inconsistent with our expectations. The notion of previous research that trading volumes would increase when investors are overconfident may be simplified. We suppose that rapid modulation in trading strategies had been adopted, which results from specialization and more information that securities firms possess. Therefore, the best strategy for investors that want to use securities firms’ prediction information would be to adjust investing strategies as price drifts are inconsistent with those predictions. This may help to avoid opposite trading to dealers and to decrease risks.

以往研究認為個體投資人普遍有過度自信現象、過度自信會呈週期性起伏，及造成低估風險、過度交易及損失。本文想了解機構投資人是否也會過度自信？其過度自信時是否也會增加交易量？本文首度利用心理學常用的自信衡量指標，以國內證券商預測上市大盤短期走勢的資料來計算自信分數，並從預測漲、跌及盤整三項分別進行，精確衡量出券商自信程度，再觀察分數的變化及其與自營部門短期交易量的關係，結果發現： 一、分成預測未來5日與10日兩種情形，樣本券商除對5日盤整的預測似無過度自信外（但這可能是預期盤整的區間過大，造成預測準確的假象所致），其餘情形皆明顯呈過度自信。又若不分三項，整體上，20家樣本券商對大盤的預測有過度自信現象。 二、將指數漲、跌大波段各分成5小段，發現中段自信分數最高，第1、4段次之，第2、5段最低，表示第3小段的大盤走勢最難預測，也顯示過度自信程度的變化會呈週期性起伏，即過去的成功會使後來的自信提高，而過度自信又會使預測誤差增加。 三、由看漲、看跌或盤整分別觀察，發現證券商過度自信是會增加某一方向的交易，但方向與預期相反，這可能是快速調整交易策略所致，造成買、賣量（以金額或股數表示）互有增減變化，而總量未必增加。 以上可使投資人了解證券商的專業與資訊優勢可能使其過度自信有所降低，因而其交易行為有很大機動性：過度自信於看漲（跌）反而少買（多買少賣），預測盤整則傾向賣出。因此，習於參考證券商盤勢解析資訊之投資人如發現行情與券商預測不符，宜即調整投資方向，以免與自營商對作，增加投資風險。 再者，若所有投資人（含證券商）都能儘量降低過度自信程度，使對大盤走勢的預測正確率提高，應可減少誤導其資訊參考者之行為，進而降低股價不必要的短期波動，使市場效率更加提升，而資產市價與真值的差距即可縮小，這將有助於企業融資政策的擬定與執行。最後，自營商高機動性的交易行為隱喻提高專業能力與市場資訊透明度是降低過度自信很好的方法。