自比特幣成為第一個發之加密貨幣至今，加密貨幣市場蓬勃發展，其高波動度所帶來之高報酬吸引投資者們趨之若鶩。波動度在金融領域中是十分重要的一個影響因子，同時機器學習在金融領域不斷的被廣泛採用，能夠提供更加理性之投資決策。本篇論文結合GARCH-LSTM(Long Short-term Memory)集成模型，選取15種加密貨幣，對其波動度進行預測，期望能夠精確預測未來波動度。Modern Portfolio Theory（MPT）存在相關性矩陣條件數（Conditional number）過高，MPT對於參數值過於敏感。Lopez de Prado於2016結合機器學習及圖論，提出Hierarchical Risk Parity（本研究以下簡稱HRP），對相關係數矩陣進行降維，期望能夠解決此問題。本文結合上述模擬出加密貨幣波動度後，採用HRP模型決定15個加密貨幣配置之權重，形成一個相較於MPT更加穩定之投資組合。最後結合VolTarget Portfolio概念，以前述15加密貨幣組成之投資組合是為風險性資產，並使用穩定幣USDT做為無風險性資產，藉由調整兩者權重設定整體投資組合波動度，期望能夠為加密貨幣建立一個更加穩定的投資策略。在準確預測未來波動度的情況下，藉由設定投資組合波動度，在享有加密貨幣高報酬情況下，亦能取得更加平穩之權益曲線。

Since Bitcoin became the first cryptocurrency to be issued, the cryptocurrency market has flourished. The higher returns brought by its high volatility compared to general assets have attracted investors. Volatility is a very important factor in the financial field, and machine learning is widely used in financial field, which can provide more rational investment decisions. This paper combines the GARCH-LSTM (Long Short-Term Memory) model to predict the fifteen selected cryptocurrency’s volatility, accurately predict their future volatility. Modern Portfolio Theory (MPT) has a high conditional number coefficient in correlation coefficient matrix. That is the reason why MPT is too sensitive to parameter values. Lopez de Prado (2016) proposes Hierarchical Risk Parity (HRP) by combining machine learning and the graph theory, which reduces the dimension of the correlation coefficient matrix, to solve this problem. After simulation the volatility of cryptocurrency based on the above method, this paper uses the HRP model to determine the weight of 15 cryptocurrencies allocations to form a more stable portfolio than MPT. In conclusion, combined with the concept of VolTarget Portfolio, the investment portfolio composed of the aforementioned 15 cryptocurrencies is a risky asset, and the stablecoin USDT is used as risk-free asset. By adjusting the weights of the two, the overall portfolio volatility is set. In the case of accurately predicting future volatility, by setting the volatility of the investment portfolio, a more stable equity curve can be obtained while enjoying the high return of cryptocurrency.

Albeverio, S., Steblovskaya, V., & Wallbaum, K. (2013). Investment instruments with volatility target mechanism. Quantitative Finance, 13(10), 1519-1528.
Albeverio, S., Steblovskaya, V., & Wallbaum, K. (2018). The volatility target effect in structured investment products with capital protection. Review of Derivatives Research, 21(2), 201-229.
Bildirici, M., & Ersin, Ö. Ö. (2013). Forecasting oil prices: Smooth transition and neural network augmented GARCH family models. Journal of Petroleum Science and Engineering, 109, 230-240.
Black, F., & Jones, R. (1987). Simplifying portfolio insurance. Journal of Portfolio Management, 14(1), 48.
Bollerslev, T. (1986). Generalized autoregressive conditional heteroskedasticity. Journal of Econometrics, 31(3), 307-327.
Bollerslev, T. (1987). A conditionally heteroskedastic time series model for speculative prices and rates of return. The Review of Economics and Statistics, 542-547.
Burggraf, T. (2021). Beyond risk parity–A machine learning-based hierarchical risk parity approach on cryptocurrencies. Finance Research Letters, 38, 101523.
De Prado, M. L. (2016). Building diversified portfolios that outperform out of sample. The Journal of Portfolio Management, 42(4), 59-69.
Di Persio, L., Garbelli, M., & Wallbaum, K. (2021). Forward-looking volatility estimation for risk-managed investment strategies during the covid-19 crisis. Risks, 9(2), 33.
Engle, R. F. (1982). Autoregressive conditional heteroscedasticity with estimates of the variance of United Kingdom inflation. Econometrica, 987-1007.
Glosten, L. R., Jagannathan, R., & Runkle, D. E. (1993). On the relation between the expected value and the volatility of the nominal excess return on stocks. The Journal of Finance, 48(5), 1779-1801.
Goel, V. K., Lazar, A. J., Warneke, C. L., Redston, M. S., & Haluska, F. G. (2006). Examination of mutations in BRAF, NRAS, and PTEN in primary cutaneous melanoma. Journal of Investigative Dermatology, 126(1), 154-160.
Hochreiter, S., & Schmidhuber, J. (1997). Long short-term memory. Neural Computation, 9(8), 1735-1780.
Hu, Y., Ni, J., & Wen, L. (2020). A hybrid deep learning approach by integrating LSTM-ANN networks with GARCH model for copper price volatility prediction. Physica A: Statistical Mechanics and its Applications, 557, 124907.
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.
Kristjanpoller, W., Fadic, A., & Minutolo, M. C. (2014). Volatility forecast using hybrid neural network models. Expert Systems with Applications, 41(5), 2437-2442.
Longin, F., & Solnik, B. (1995). Is the correlation in international equity returns constant: 1960–1990? Journal of International Money and Finance, 14(1), 3-26.
Qiu, J., Wang, B., & Zhou, C. (2020). Forecasting stock prices with long-short term memory neural network based on attention mechanism. PloS one, 15(1), e0227222.
Rabemananjara, R., & Zakoian, J.-M. (1993). Threshold ARCH models and asymmetries in volatility. Journal of applied econometrics, 8(1), 31-49.
Tseng, C.-H., Cheng, S.-T., Wang, Y.-H., & Peng, J.-T. (2008). Artificial neural network model of the hybrid EGARCH volatility of the Taiwan stock index option prices. Physica A: Statistical Mechanics and its Applications, 387(13), 3192-3200.