| Summary: | 碩士 === 東吳大學 === 國際經營與貿易學系 === 104 === This study uses the predictive value of each broker to establish four integrated prediction models so as to predict the net income of TSMC(2330) and MediaTek (2454) in 2015, and takes three error analysis indicators (MAW, RMSE and ME) to examine each broker analyst and the four integrated predictive value as well as the “Unchanged” and “The Newest” prediction performances. Finally, this study proceeds in the exploration whether the different predictive values can make up the information gap of the “window period of the earnings report” so as to be provided for the public’s reference.
As shown in the empirical results, the brokers’ predictions of the different companies are not completely stable, but influenced by the prediction frequency, brokerage nature and prediction point; the situations of consistent overvaluation/undervaluation have been presented. Therefore, the integrated prediction model can reduce the gaps of the prediction information of each of the brokers. Also, during the window period of the earnings report, the integrated prediction model can effectively predict the annual report of the listed company. Besides, “The Unchanged” has achieved good predictive performances of each year, yet was ineffective in the prediction during the window period of the earnings report. “The Newest” has been very accurate in the prediction of the years; however, for the prediction of the following year from the previous year, “The Newest” was not the best prediction method. As for the performances of the four integrated prediction model for different companies and different prediction period, there were no consistent performances, showing that the weighted integrated prediction model was not significantly better than the simple integrated prediction model.
Keywords: Financial projections, Analysts’ predict, Integrated prediction model, Prediction performance rating, Accuracy
|
|---|
