隨著時代演進，金融商品日新月異，投資者往往追求高槓桿高報酬的商品，近期衍生性金融商品-台股期貨(TX)也蓬勃發展，成為國內投資最活絡的交易工具之一。然而，隨伴而來更是高難度的風險控管問題，近年來正是人工神經網路日益突破的時代，人工神經網路概念被大量運用在各種金融商品的研究中。
    本文利用2016年1月4日至2019年9月9日間在台灣期貨交易所發行之台股期貨，應用其每日歷史報酬率，分別搭配三種波動率預測模型，GARCH、GJR-GARCH與倒傳遞類神經網絡(BPNN)，對台股期貨進行波動率預測，並採取2018年1月2日至2019年9月9日之每十分鐘報酬率作進一步的預測準確性作檢定，分別比較使用日資料與日內資料的績效預測差異。本研究績效評價指標為泰爾不等係數(Theil’s U)、均方根誤差(RMSE)、絕對平均誤差(MAE)與平均絕對誤差百分比(MAPE)進行評估。
    本文研究結果顯示，每日報酬率之日資料，其得出的預測結果較為準確，並運用傳統的GARCH模型其預測性能較佳；而在日資料中，加入倒傳遞類神經網絡(BPNN)相較於傳統的預測模型，的確可以提高其波動性的預測能力，且較能補捉台股期貨之上漲與下跌趨勢。

With the development of the times, financial products change with each passing day. Recently, derivative financial commodities-,ie. Taiwan Futures Index (TX) have also flourished and become one of the most active trading tools for domestic investment. However, accompanying it is a more difficult risk control problem. In recent years, the era of artificial neural networks has been breaking through. The concept of artificial neural networks has been widely used in the research of various financial commodities.
    This research uses Taiwan Futures Index , data period is obtained from January 4, 2016 to September 9, 2019, which were issued on the Taiwan Futures Exchange. We calculated the daily historical rate of return and used the three volatility prediction models of GARCH, GJR-GARCH, and BPNN to forecast the volatility of Taiwan Futures Index.
   Next, we took a 10-minute return from January 2, 2018 to September 9, 2019 for further prediction accuracy verification, and compared the performance prediction differences between the daily data and the high frequency data. Finally, in this study, Theil's U, RMSE, MAE, and MAPE were used as performance evaluation indicators to evaluate the prediction accuracy. The research results show that the daily data of the daily rate of return has a more accurate prediction result, and the traditional GARCH model has better prediction performance. In addition, compared with traditional forecasting models, adding BPNN can indeed improve its forecasting ability of volatility. And it can better catch up the fluctuation trend of Taiwan stock futures.

目錄  
目錄..................................................iv 
圖目錄................................................vi 
表目錄................................................vi

第一章 緒論 1
第一節 研究背景與動機............1
第二節 研究目的.................3
第三節 研究架構與流程............4

第二章 理論與相關文獻 6
第一節 高頻數據的相關文獻...........6
第二節 波動率估計模型的相關文獻.....7
第三節 波動率預測模型的相關文獻.....9

第三章 研究方法 12
第一節 研究資料定義與來源........12
第二節 波動率估計模型-真實波動率RV、歷史波動率HV定義與概念....13
第三節 波動率預測模型-GARCH、GJR-GARCH與BPNN模型定義與概念...14
第四節 績效評估模型.....18

第四章　實證結果與分析	 20
第一節 敘述統計量分析.........20
第二節 資料檢定...............25
第三節 GARCH模型預測分析.......27
第四節 GJR-GARCH模型預測分析...33
第五節 BPNN模型預測分析........38
第六節 績效檢定結果............43

第五章　結論與建議 50
第一節 結論.....50
第二節 建議.....51
參考文獻........52

 
表目錄
表 1台股期貨之基本敘述統計量....21
表 2單根檢定結果.........25
表 3異質性檢定結果.......26
表 4 GARCH估計檢定結果...28
表 5 GARCH預測結果之敘述統計量...30
表 6 GJR-GARCH估計檢定結果......34
表 7 GJR-GARCH預測檢定結果之敘述統計量.......36
表 8 倒傳遞類神經網路(BPNN)參數設定..........39
表 9 BPNN預測檢定結果之敘述統計量............40
表 10每十分鐘報酬率之GARCH與GJR-GARCH預測模型之
     績效評估結果....43
表 11每日報酬率之GARCH、GJR-GARCH與BPNN預測模型之
     績效評估結果....45
表 12各模型績效比較總表.......................49

圖目錄
圖 1 研究架構流程圖.....5
圖 2 倒傳遞類神經網路架構圖......17
圖 3 每十分鐘報酬率時間序列圖....22
圖 4 每日報酬率時間序列圖........22
圖 5 每十分鐘真實報酬波動性走勢圖...23
圖 6 每日真實報酬波動性走勢圖.......23
圖 7 不同資料型態之次數分配表.......24
圖 8 每日報酬波動性GARCH檢定之殘差圖....29
圖 9 各預測報酬波動性於GARCH預測模型之次數分配表.........31
圖 10每十分鐘報酬波動性預測:GARCH預測模型-台股期貨(TX)...32
圖 11每日報酬波動性預測:GARCH預測模型-台股期貨(TX)......32
圖 12每日報酬波動性GJR-GARCH檢定之殘差圖.....35
圖 13各預測報酬波動性於GJR-GARCH預測模型之次數分配表....36
圖 14每十分鐘報酬波動性預測: GJR-GARCH預測模型-台股期貨(TX)...37
圖 15每日報酬波動性預測: GJR-GARCH預測模型-台股期貨(TX)...38
圖 16倒傳遞類神經網路(BPNN)結構圖-1....39
圖 17倒傳遞類神經網路(BPNN)結構圖-2....40
圖 18各預測報酬波動性於BPNN預測模型之次數分配表......41
圖 19訓練資料-每日報酬波動性預測:BPNN預測模型-台股期貨(TX)..42
圖 20測試資料-每日報酬波動性預測:BPNN預測模型-台股期貨(TX)..42
圖 21每十分鐘GARCH預測報酬波動性與真實波動性之走勢圖........44
圖 22每十分鐘GJR-GARCH預測報酬波動性與真實波動性之走勢圖....44
圖 23每日GARCH預測報酬波動性與真實報酬性之走勢圖..........46
圖 24每日GJR-GARCH預測報酬波動性與真實報酬性之走勢圖......46
圖 25訓練資料-每日BPNN預測報酬波動性與真實波動性之走勢圖...47
圖 26測試資料-每日BPNN預測報酬波動性與真實波動性之走勢圖...47
圖 27 BPNN每日預測報酬波動性與真實報酬性之走勢圖..........48

參考文獻 
一、 中文文獻 
1. 張育維(2011)。改良式類神經網路預測模式於股價預測之研究。北商學報，
第 23 期，頁 1-18。 
2. 梁雪富(2018)。GARCH 家族模型對涉險值(VaR)估計的績效評估。臺灣銀行
季刊，第 69 卷，第 1 期，頁 90-102。 
3. 廖偉真、雷立芬(2010)。不同樣本頻率之股價波動性估計-GARCH.TGARCH
與 EGARCH 之比較。台灣銀行季刊，第 61 卷，第 4 期，頁 294-307。 
 
二、 英文文獻 
1. Andersen ,T. G. and T. Bollerslev (1998), “ Answering the Skeptics: Yes, Standard 
Volatility Models do Provide Accurate Forecasts,” International Economic Review, 
Vol.39, No.4 pp. 885-905 
2. Andersen ,T. G.T.Bollerslev,F. X Diebold and P. Labys (2001) ,“The Distribution 
of Realized Exchange Rate Volatility,”Journal of the American Statistical 
Association, Vol. 96, No. 453., pp.42-45 
3. Andersen, T.G, T. Bollerslev.,F. X. Diebold, and P. Labys (2003), “Modeling and 
Forecasting Realized Volatility,” Econometrica .Vol. 71,pp. 529-626. 
4. Andersen,T.G, T. Bollerslev,and N. Meddahi (2003), “Correcting The Errors: 
Volatility Forecast Evaluation Using High-Frequency Data and Realized 
Volatilities,” Econometrica, Vol. 73, No. 1 , 279–296 
5. Andersen ,T. G., T.Bollerslev, P. Christoffersen and F. X.Diebold (2007), “A 
Framework for Exploring the Macroeconomic Determinants of Systematic Risk,” 
Center for Financial Studies. 
6. Adebiyi, A. A. , . A. O. Adewumi and C. K. Ayo (2014) , “ Comparison of ARIMA 
and Artificial Neural Networks Models for Stock Price Prediction,” Journal of 
Applied Mathematics, Vol. 2, No. 1, pp.1-7. 
7. Bollerslev ,T.(1986),“Generalized Autoregressive Conditional Heteroskedasticity,” 
Journal of Econometrics ,Vol.31,pp. 307-327. 
8. Bin Zhou (1996), “High-Frequency Data and Volatility in Foreign-Exchange Rates,” 
Journal of Business & Economic Statistics, Vol. 14, issue 1, pp.45-52 
9. Barndorff-Nielsen, O.E.  and N. Shephard (1998), “Econometric Analysis of Realized Volatility and Its Use In Estimating Stochastic Volatility Models,” 
Journal of the Royal Statistical Society Series B. 
10. Blair, B., Poon, S. H., and Taylor, S. J. (2001), “Forecasting S&P 100 Volatility: 
The Incremental Information Content of Implied Volatilities and High Frequency 
Index Returns,”  Journal of Econometrics, Vol. 105, No. 1, pp.5-26. 
11. Christopher G Lamoureux and William Lastrapes (1993), “ Forecasting Stock
Return Variance: Toward an Understanding of Stochastic Implied Volatilities,” 
Review of Financial Studies, Vol. 6, Issue 2, pp.293-326. 
12. Dickey, D. and W.A. Fuller (1981), “ Likelihood ratio statistics for autoregressive 
time series with a unit root,”  Econometric, Vol. 49, No.4, pp.1057-1072. 
13. Drost, F. C. and T. E. Nijman (1993), “Temporal Aggregation of Garch Processes,” 
Econometrics, Vol. 61, No.4, pp.909-927. 
14. Engle, R. F. (1982), “ Autoregressive Conditional Heteroscedasticity with 
Estimates of the Variance of United Kingdom Inflation,” Econometrica, Vol. 50, 
No. 4, pp. 987-1007. 
15. Fama, E. F.(1965), “The Behavior of Stock-Market Prices,” The Journal of 
Business, Vol. 38, No. 1. pp. 34-105. 
16. Glosten, L. R., R. Jagannathan and D, E, Runkle (1993), “ On the Relation between 
the Expected Value and the Volatility of the Nominal Excess Return on Stocks,” 
The Journal of Finance, Vol. 48, No. 5, pp. 1779-1801 
17. Gwilym,O. A. and M. Buckle (1999), “ The Lead-Lag Relationship between the 
FTSE100 Stock Index and Its Derivative Contracts,” Financial Economics, Vol. 11, 
Issue.4. 
18. Hornik,K. M. Stinchcombe. and H. White (1989), “ Multilayer Feedforward 
Networks Are Universal Approximators,” Elsevier Ltd. Neural Networks, Vol. 2, 
Issue. 5, pp.359-366. 
19. J.P. Morgan and Reuters (1996), “RiskMetrics －Technical Document,” J.P. 
Morgan, 4th ed. 
20. Hopfield, J.(1984), “Neurons with Graded Response Have Collective 
Computational Properties like Those of Two-State Neurons,”
 
Proceedings of the 
National Academy of Sciences, Vol. 81, pp. 3088-3092. 
21. Hutchinson; J. M. A.W. Lo and T. Poggio (1994), “ A Nonparametric Approach to 
Pricing and Hedging Derivative Securities Via Learning Networks,” The Journal of 
Finance, Vol. 49, No. 3, Papers and Proceedings Fifty-Fourth Annual Meeting of 
the American Finance Association, Boston, Massachusetts, pp. 851-889. 
22. Lamoureux. C. G. and W. D. Lastrapes (1993), “ Forecasting Stock-Return 
Variance: Toward an Understanding of Stochastic Implied Volatilities,” Review of 
Financial Studies, Vol. 6, Issue 2, pp.293-326. 
23. Lua,,X.-F. D. Quea, and G. Caoa (2016), “ Volatility Forecast Based on the Hybrid 
Artificial Neural Network and GARCH-Type Models,” Procedia Computer 
Science, Vol. 91, pp.1044-1049. 
24. Mandelbrot, B. (1963), “The Variation of Certain Speculative Prices,” The Journal 
of Business, Vol. 36, No. 4, pp. 394-419 
25. Merton, R. C.(1980), “ On Estimating The Expected Return On The Market An 
Exploratory Investigation,” Journal of Financial Economics, Vol.8, pp.323-361. 
26. Nelson , D. B.(1991) , “ Conditional Heteroskedasticity in Asset Returns: A New 
Approach,” Econometrics, Vol. 59, No. 2, pp.347-370. 
27. Tomáš Kˇ rehlík and Jozef  Baruník (2016), “ Combining High Frequency Data 
With Non-Linear Models For Forecasting Energy Market Volatility,” Expert 
Systems with Applications, Vol. 55, pp.222-242.