論文使用權限 Thesis access permission:校內校外完全公開 unrestricted
開放時間 Available:
校內 Campus: 已公開 available
校外 Off-campus: 已公開 available
論文名稱 Title |
比較ARMA-GARCH、Prophet、LSTM模型預測加密貨幣波動度之準確率 Predicting the accuracy of cryptocurrency volatility with ARMA-GARCH, Prophet and LSTM models |
||
系所名稱 Department |
|||
畢業學年期 Year, semester |
語文別 Language |
||
學位類別 Degree |
頁數 Number of pages |
124 |
|
研究生 Author |
|||
指導教授 Advisor |
|||
召集委員 Convenor |
|||
口試委員 Advisory Committee |
|||
口試日期 Date of Exam |
2023-06-13 |
繳交日期 Date of Submission |
2024-03-03 |
關鍵字 Keywords |
加密貨幣、實際波動度(Realized Volatility, RV)、ARMA-GARCH模型、Prophet模型、RNN-LSTM模型 Cryptocurrency, Realized Volatility (RV), ARMA-GARCH model, Prophet model, RNN-LSTM model |
||
統計 Statistics |
本論文已被瀏覽 188 次,被下載 15 次 The thesis/dissertation has been browsed 188 times, has been downloaded 15 times. |
中文摘要 |
近年來,科技爆炸性的成長促使全球的金融市場邁向全面數位化,現今的技術允許人們使用加密貨幣進行投資或消費,為現代熱門的金融資產之一;截至2022年11月,根據幣安(Binance)統計,成交量排行前三名的幣種為:比特幣(BTC)、以太幣(ETH)和幣安幣(BNB)。但加密貨幣市場價格存在高波動性、非常態分佈、長尾以及存在極端事件的特性,投資人可能會在短時間迅速賺取或損失大量資金。為了避免過度損失,透過預測波動度可以提供投資人在做投資決策時重要的參考依據。 故本研究以比特幣(BTC)、以太幣(ETH)和幣安幣(BNB)為研究對象,使用3分鐘、5分鐘、15分鐘之歷史價格建構每日實際波動度(Realized Volatility, RV),分別比較ARMA-GARCH、Prophet以及RNN-LSTM模型在滑動窗口算法(Sliding Window Algorithm)預測下一日、下兩日、下三日及下五日的實際波動度(Realized Volatility, RV)之準確率。研究結果顯示總體而言RNN-LSTM擁有較佳的預測能力。 |
Abstract |
In recent years, the explosive growth of technology has prompted the global financial market to move towards full digitization. Today's technology allows people to use cryptocurrencies for investment or consumption, making them one of the most popular financial assets in modern times. As of November 2022, according to Binance, the top three currencies by trading volume are Bitcoin (BTC), Ethereum (ETH), and Binance Coin (BNB). However, due to the characteristics of high volatility, abnormal distribution, long tail, and extreme events in the cryptocurrency market, investors may quickly earn or lose a large amount of money in a short period of time. To avoid excessive losses, predicting volatility can provide investors with an important reference when making investment decisions. Therefore, this study focuses on Bitcoin (BTC), Ethereum (ETH), and Binance Coin (BNB) as the research objects. It utilizes historical prices at 3-minute, 5-minute, and 15-minute intervals to construct daily Realized Volatility (RV). The study compares the accuracy of three models: ARMA-GARCH, Prophet, and RNN-LSTM, implemented using the sliding window algorithm. These models aim to predict the Realized Volatility (RV) for the next one, two, three, and five days. The research results demonstrate that RNN-LSTM generally exhibits better predictive ability. |
目次 Table of Contents |
論文審定書 i 誌謝 ii 摘要 iii Abstract iv 目錄 v 圖目錄 viii 表目錄 xiv 第一章 緒論 1 1.1 研究背景與動機 1 1.2 研究目的 4 1.3 研究架構 5 第二章 文獻回顧 5 第三章 研究方法與模型介紹 9 3.1 資料預處理 9 3.1-1 實際波動度(Realized Volatility, RV) 9 3.1-2滑動窗口算法(Sliding Window Algorithm) 9 3.1-3單根檢定(Augmented Dickey-Fuller, ADF) 10 3.1-4 決定落後期數(Lag) 11 3.2 建立預測模型 12 3.2-1 ARMA-GARCH 12 3.2.2 Prophet 13 3.2.3 RNN-LSTM 15 3.3 模型評估指標 19 第四章 實證結果與分析 21 4.1 資料來源與資料前處理 21 4.1-1 ADF檢定結果 23 4.1-2最適落後期 23 4.2 實證結果 24 4.2-1 ARMA-GARCH (1,1) 24 4.2-2 Prophet 27 4.2-3 RNN-LSTM 33 4.2-4 模型比較 44 參考文獻 50 附錄 53 1. 三層LSTM模型準確率比較表 53 2. ARMA-GARCH預測表現圖 54 i.比特幣(BTC) 54 ii.以太幣(ETH) 60 iii.幣安幣(BNB) 66 3. Prophet預測表現圖 73 i.比特幣(BTC) 73 ii.以太幣(ETH) 79 iii.幣安幣(BNB) 85 4. RNN-LSTM預測表現圖 91 i.比特幣(BTC) 91 ii.以太幣(ETH) 97 iii.幣安幣(BNB) 103 圖目錄 圖 1 研究架構圖 5 圖 2 滑動窗口算法(Sliding Window Algorithm)架構 10 圖 3 LSTM模型每層記憶區塊(memory block)架構圖 16 圖 4 tanh函數示意圖 17 圖 5 sigmoid函數示意圖 18 圖 6 2017/08/18~2022/11/27每日收盤價走勢圖 22 圖 7 2017/08/18~2022/11/27 每日收益率 22 圖 8 Prophet樣本內資料趨勢圖(BTC) 28 圖 9 Prophet樣本內資料趨勢圖(ETH) 29 圖 10 Prophet樣本內資料趨勢圖(BNB) 30 圖 11 LSTM一層記憶區塊(memory block)示意圖 34 圖 12 ARMA-GARCH下一期預測圖(BTC) 55 圖 13 ARMA-GARCH下兩期預測圖(BTC) 55 圖 14 ARMA-GARCH下三期預測圖(BTC) 56 圖 15 ARMA-GARCH下五期預測圖(BTC) 56 圖 16 ARMA-GARCH下一期預測圖(BTC) 57 圖 17 ARMA-GARCH下兩期預測圖(BTC) 57 圖 18 ARMA-GARCH下三期預測圖(BTC) 58 圖 19 ARMA-GARCH下五期預測圖(BTC) 58 圖 20 ARMA-GARCH下一期預測圖(BTC) 59 圖 21 ARMA-GARCH下兩期預測圖(BTC) 59 圖 22 ARMA-GARCH下三期預測圖(BTC) 60 圖 23 ARMA-GARCH下五期預測圖(BTC) 60 圖 24 ARMA-GARCH下一期預測圖(ETH) 61 圖 25 ARMA-GARCH下兩期預測圖(ETH) 61 圖 26 ARMA-GARCH下三期預測圖(ETH) 62 圖 27 ARMA-GARCH下五期預測圖(ETH) 62 圖 28 ARMA-GARCH下ㄧ期預測圖(ETH) 63 圖 29 ARMA-GARCH下兩期預測圖(ETH) 63 圖 30 ARMA-GARCH下三期預測圖(ETH) 64 圖 31 ARMA-GARCH下五期預測圖(ETH) 64 圖 32 ARMA-GARCH下一期預測圖(ETH) 65 圖 33 ARMA-GARCH下兩期預測圖(ETH) 65 圖 34 ARMA-GARCH下三期預測圖(ETH) 66 圖 35 ARMA-GARCH下五期預測圖(ETH) 66 圖 36 ARMA-GARCH下一期預測圖(BNB) 67 圖 37 ARMA-GARCH下兩期預測圖(BNB) 67 圖 38 ARMA-GARCH下三期預測圖(BNB) 68 圖 39 ARMA-GARCH下五期預測圖(BNB) 68 圖 40 ARMA-GARCH下一期預測圖(BNB) 69 圖 41 ARMA-GARCH下兩期預測圖(BNB) 69 圖 42 ARMA-GARCH下三期預測圖(BNB) 70 圖 43 ARMA-GARCH下五期預測圖(BNB) 70 圖 44 ARMA-GARCH下一期預測圖(BNB) 71 圖 45 ARMA-GARCH下兩期預測圖(BNB) 71 圖 46 ARMA-GARCH下三期預測圖(BNB) 72 圖 47 ARMA-GARCH下五期預測圖(BNB) 72 圖 48 Prophet下ㄧ期預測圖(BTC) 73 圖 49 Prophet下兩期預測圖(BTC) 74 圖 50 Prophet下三期預測圖(BTC) 74 圖 51 Prophet下五期預測圖(BTC) 75 圖 52 Prophet下一期預測圖(BTC) 75 圖 53 Prophet下兩期預測圖(BTC) 76 圖 54 Prophet下三期預測圖(BTC) 76 圖 55 Prophet下五期預測圖(BTC) 77 圖 56 Prophet下一期預測圖(BTC) 77 圖 57 Prophet下兩期預測圖(BTC) 78 圖 58 Prophet下三期預測圖(BTC) 78 圖 59 Prophet下五期預測圖(BTC) 79 圖 60 Prophet下一期預測圖(ETH) 80 圖 61 Prophet下兩期預測圖(ETH) 80 圖 62 Prophet下三期預測圖(ETH) 80 圖 63 Prophet下五期預測圖(ETH) 81 圖 64 Prophet下一期預測圖(ETH) 82 圖 65 Prophet下兩期預測圖(ETH) 82 圖 66 Prophet下三期預測圖(ETH) 82 圖 67 Prophet下五期預測圖(ETH) 83 圖 68 Prophet下一期預測圖(ETH) 83 圖 69 Prophet下兩期預測圖(ETH) 84 圖 70 Prophet下三期預測圖(ETH) 84 圖 71 Prophet下五期預測圖(ETH) 85 圖 72 Prophet下一期預測圖(BNB) 86 圖 73 Prophet下兩期預測圖(BNB) 86 圖 74 Prophet下三期預測圖(BNB) 86 圖 75 Prophet下五期預測圖(BNB) 87 圖 76 Prophet下一期預測圖(BNB) 87 圖 77 Prophet下兩期預測圖(BNB) 88 圖 78 Prophet下三期預測圖(BNB) 88 圖 79 Prophet下五期預測圖(BNB) 89 圖 80 Prophet下一期預測圖(BNB) 89 圖 81 Prophet下兩期預測圖(BNB) 90 圖 82 Prophet下三期預測圖(BNB) 90 圖 83 Prophet下五期預測圖(BNB) 91 圖 84 RNN-LSTM下一期預測圖(BTC) 92 圖 85 RNN-LSTM下兩期預測圖(BTC) 92 圖 86 RNN-LSTM下三期預測圖(BTC) 93 圖 87 RNN-LSTM下五期預測圖(BTC) 93 圖 88 RNN-LSTM下ㄧ期預測圖(BTC) 94 圖 89 RNN-LSTM下兩期預測圖(BTC) 94 圖 90 RNN-LSTM下三期預測圖(BTC) 95 圖 91 RNN-LSTM下五期預測圖(BTC) 95 圖 92 RNN-LSTM下一期預測圖(BTC) 96 圖 93 RNN-LSTM下兩期預測圖(BTC) 96 圖 94 RNN-LSTM下三期預測圖(BTC) 97 圖 95 RNN-LSTM下五期預測圖(BTC) 97 圖 96 RNN-LSTM下一期預測圖(ETH) 98 圖 97 RNN-LSTM下兩期預測圖(ETH) 98 圖 98 RNN-LSTM下三期預測圖(ETH) 99 圖 99 RNN-LSTM下五期預測圖(ETH) 99 圖 100 RNN-LSTM下一期預測圖(ETH) 100 圖 101 RNN-LSTM下兩期預測圖(ETH) 100 圖 102 RNN-LSTM下三期預測圖(ETH) 101 圖 103 RNN-LSTM下五期預測圖(ETH) 101 圖 104 RNN-LSTM下一期預測圖(ETH) 102 圖 105 RNN-LSTM下兩期預測圖(ETH) 102 圖 106 RNN-LSTM下三期預測圖(ETH) 103 圖 107 RNN-LSTM下五期預測圖(ETH) 103 圖 108 RNN-LSTM下一期預測圖(BNB) 104 圖 109 RNN-LSTM下兩期預測圖(BNB) 104 圖 110 RNN-LSTM下三期預測圖(BNB) 105 圖 111 RNN-LSTM下五期預測圖(BNB) 105 圖 112 RNN-LSTM下ㄧ期預測圖(BNB) 106 圖 113 RNN-LSTM下兩期預測圖(BNB) 106 圖 114 RNN-LSTM下三期預測圖(BNB) 107 圖 115 RNN-LSTM下五期預測圖(BNB) 107 圖 116 RNN-LSTM下一期預測圖(BNB) 108 圖 117 RNN-LSTM下兩期預測圖(BNB) 108 圖 118 RNN-LSTM下三期預測圖(BNB) 109 圖 119 RNN-LSTM下五期預測圖(BNB) 109 表目錄 表 1 日收益率敘述統計表 23 表 2 ADF檢定表 23 表 3 AIC值比較 24 表 4 ARMA(0,1)-GARCH(1,1)每期預測結果(BTC) 25 表 5 ARMA(0,1)-GARCH(1,1)每期預測結果(ETH) 26 表 6 ARMA(1,0)-GARCH(1,1)每期預測結果(BNB) 27 表 7 Prophet每期預測結果(BTC) 31 表 8 Prophet每期預測結果(ETH) 32 表 9 Prophet每期預測結果(BNB) 33 表 10 LSTM模型建立及參數數量(BTC) 37 表 11 RNN-LSTM每期預測結果(BTC) 39 表 12 LSTM模型建立及參數數量(ETH) 39 表 13 RNN-LSTM每期預測結果(ETH) 41 表 14 LSTM模型建立及參數數量(BNB) 42 表 15 RNN-LSTM每期預測結果(BNB) 44 表 16 各模型波動度預測誤差比較表(BTC) 44 表 17 各模型波動度預測誤差比較表(ETH) 45 表 18 各模型波動度預測誤差比較表(BNB) 47 表 19 三層LSTM模型準確率比較表 53 |
參考文獻 References |
Aker, Y. (2022). Analysis of Price Volatility in BIST 100 Index With TimeSeries:Comparison of Fbprophet and LSTM Model. European Journal of Science and Technology, 35, 89-93. https:// doi.org/10.31590/ejosat.1066722. Akaike, H. (1974). A new look at the statistical model identification. IEEE transactions on automatic control, 19(6), 716-723. A. Chatterjee, H. Bhowmick, and J. Sen, “Stock volatility prediction using time series and deep learning approach” Proceedings of the 2022 IEEE 2nd Mysore Sub Section International Conference (MysuruCon’22), pp 1-6, October 16-17, 2022. DOI: 10.1109/MysuruCon55714.2022.9972559 Balaban, E., Bayar, A., Faff, R.W., 2006. Forecasting stock market volatility: Further international evidence. The European Journal of Finance 12, 171-188. Bollerslev, T. (1986). Generalized autoregressive conditional heteroskedasticity. Journal of econometrics, 31(3), 307-327. Borneklint, N. (2021). Forecasting prices of Bitcoin and Google stock with ARIMA vs Face- book Prophet. Master Thesis. University West, Sweden. Bowden J, Gemayel R (2022) Sentiment and trading decisions in an ambiguous environment: a study on cryptocurrency traders. J Int Financ Mark Inst Money 80:101622. Delfabbro P, King DL, Williams J (2021) The psychology of cryptocurrency trading: Risk and protective factors. J Behav Addict Dickey, D. A., & Fuller, W. A. (1979). Distribution of the estimators for autoregressive time series with a unit root. Journal of the American statistical association, 74(366a), 427-431. Eranga Bandara, Xueping Liang, Peter Foytik, Sachin Shetty, Crissie Hall, Daniel Bowden, Nalin Ranasinghe, and Kasun De Zoysa. 2021. A blockchain empowered and privacy preserving digital contact tracing platform. Information processing & management 58, 4 (2021), 102572. Engle, R. F. (1982). Autoregressive conditional heteroscedasticity with estimates of the variance of United Kingdom inflation. Econometrica: Journal of the econometric society, 987-1007. Engle, R. F., & Yoo, B. S. (1987). Forecasting and testing in co-integrated systems. Journal of econometrics, 35(1), 143-159. Gbadebo, Adedeji D., Ahmed O. Adekunle, Adedokun Wole, Adebayo-Oke A. Lukman, and Joseph Akande. 2021. BTC price volatility: Fundamentals versus information. Cogent Business & Management 8: 1. Hastie, T. & Tibshirani, R. (1987), ‘Generalized additive models: some applications’, Jour- nal of the American Statistical Association 82(398), 371–386. Harvey, A. & Peters, S. (1990), ‘Estimation procedures for structural time series models’, Journal of Forecasting 9, 89–108. Hochreiter, S., & Schmidhuber, J. (1997). Long short-term memory. Neural computation, 9(8), 1735-1780. Iqbal, M., Iqbal, M., Jaskani, F., Iqbal, K., & Hassan, A. (2021). Time-series prediction of cryptocurrency market using machine learning techniques. EAI Endorsed Transactions on Creative Technologies, 8(28). Kim, H. Y., & Won, C. H. (2018). Forecasting the volatility of stock price index: A hybrid model integrating LSTM with multiple GARCH-type models. Expert Systems with Applications, 103, 25-37. King, T., & Koutmos, D. (2021). Herding and feedback trading in cryptocurrency markets. Annals of Operations Research, 300, 79-96. Koutmos, D. (2023). Investor sentiment and bitcoin prices. Review of Quantitative Finance and Accounting, 60(1), 1-29. Kuhe, D. A. (2019). Modeling volatility of agricultural commodity food price index in Nigeria using ARMA-GARCH models. Mediterranean Journal of Basic and Applied Sciences, 3(1), 55-75. Kuikka, O. (2019). Can cryptocurrency come to fulfil the functions of money? An evaluation of cryptocurrency as a global currency. Kuznietsova, N., Bidyuk, P., & Kuznietsova, M. (2022). Data mining methods, models and solutions for Big Data cases in telecommunication industry. Lecture Notes in Computational Intelligence and Decision Making: 2021 International Scientific Conference" Intellectual Systems of Decision-making and Problems of Computational Intelligence”, Proceedings, Liu, R., Jiang, Y., & Lin, J. (2022). Forecasting the Volatility of Specifc Risk for Stocks with LSTM. Procedia Computer Science, 202, 111-114. N. Kumar, R. Chauhan and G. Dubey, "Forecasting of Stock Price Using LSTM and Prophet Algorithm", Applications of Artificial Intelligence and Machine Learning, pp. 141-155, 2021. Saikkonen, P. (1992). Estimation and testing of cointegrated systems by an autoregressive approximation. Econometric theory, 8(1), 1-27. Schwarz, G. (1978). Estimating the dimension of a model. The annals of statistics, 461-464. Shen, Z., Wan, Q., & Leatham, D. J. (2021). Bitcoin return volatility forecasting: A comparative study between GARCH and RNN. Journal of Risk and Financial Management, 14(7), 337. Solis, F. J., & Wets, R. J.-B. (1981). Minimization by random search techniques. Mathematics of operations research, 6(1), 19-30. Taylor, S. J., & Letham, B. (2018). Forecasting at scale. The American Statistician, 72(1), 37-45. Tsay, R. S. (1984). Order selection in nonstationary autoregressive models. The annals of statistics, 12(4), 1425-1433. Yenidogan I, Cayir A, Kozan O, Dag T, Arslan C. Bitcoin Forecasting Using ARIMA and PROPHET. In 2018 3rd International Conference on Computer Science and Engineering (UBMK); 2018 Sep: IEEE. Zahid, M., Iqbal, F., Raziq, A., & Sheikh, N. (2022). Modeling and Forecasting the Realized Volatility of Bitcoin using Realized HAR-GARCH-type Models with Jumps and Inverse Leverage Effect. Sains Malaysiana, 51(3), 929-942. |
電子全文 Fulltext |
本電子全文僅授權使用者為學術研究之目的,進行個人非營利性質之檢索、閱讀、列印。請遵守中華民國著作權法之相關規定,切勿任意重製、散佈、改作、轉貼、播送,以免觸法。 論文使用權限 Thesis access permission:校內校外完全公開 unrestricted 開放時間 Available: 校內 Campus: 已公開 available 校外 Off-campus: 已公開 available |
紙本論文 Printed copies |
紙本論文的公開資訊在102學年度以後相對較為完整。如果需要查詢101學年度以前的紙本論文公開資訊,請聯繫圖資處紙本論文服務櫃台。如有不便之處敬請見諒。 開放時間 available 已公開 available |
QR Code |