請用此 Handle URI 來引用此文件：
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/56672完整後設資料紀錄
| DC 欄位 | 值 | 語言 |
|---|---|---|
| dc.contributor.advisor | 陳和麟(Ho-Lin Chen) | |
| dc.contributor.author | Fu-Min Szu | en |
| dc.contributor.author | 司福民 | zh_TW |
| dc.date.accessioned | 2021-06-16T05:41:17Z | - |
| dc.date.available | 2025-07-06 | |
| dc.date.copyright | 2020-08-14 | |
| dc.date.issued | 2020 | |
| dc.date.submitted | 2020-07-27 | |
| dc.identifier.citation | [1] J. H. Holland, “Genetic Algorithms Computer,” Scientific American, vol. 267, no. 1, pp. 66–73, 1992. vii, 2 [2] J. B. Heaton, N. G. Polson, and J. H. Witte, “Deep learning for finance: deep portfolios,” Applied Stochastic Models in Business and Industry, vol. 33, no. 1, pp. 3–12, 2017. vii [3] H. J. Shin and J. Ryu, “A dynamic hedging strategy for option transaction using artificial neural networks,” International Journal of Software Engineering and its Applications, vol. 6, no. 4, pp. 111–116, 2012. vii, 47 [4] A. Loreggia, Y. Malitsky, H. Samulowitz, and V. Saraswat, “Deep learning for algorithm portfolios,” in 30th AAAI Conference on Artificial Intelligence, AAAI 2016, 2016, pp. 1280–1286. vii, 1 [5] N. W. Yang, “The Empirical Research In The Application of Stock Selection Strategy In CPPI,” Master’s thesis, National Taiwan University, 2008. vii [6] M. C. Jensen, “The Performance of Mutual Funds in the Period 1945-1964,” The Journal of Finance, vol. 23, no. 2, pp. 389–416, 1968. vii, 38 [7] F. A. Sortino, R. v. d. Meer, and A. Plantinga, “The Dutch Triangle,” The Journal of Portfolio Management, vol. 26, no. 1, pp. 50–57, 1999. vii, 37 [8] S. Williams, D. Fenn, and M. Mcdonald, “Risk On ‘‘ Risk Off Fixing a broken investment process,” HSBC Global Research, no. April, 2012. vii, 17 [9] T. J. Moskowitz, Y. Hua, and L. Heje, “Time series momentum $,” Journal of Financial Economics, vol. 104, no. 2, pp. 228–250, 2012. vii [10] L. Yu, S. Wang, and K. K. Lai, “Neural network-based mean-variance- skewness model for portfolio selection,” Computers and Operations Research, vol. 35, no. 1, pp. 34–46, 2008. vii, viii, 47 [11] H. Markowitz, “Portfolio Selection,” The Journal of Finance, vol. 7, no. 1, pp. 77–91, 1952. vii, 1, 23 [12] C. R. Harvey and A. Siddique, “Conditional skewness in asset pricing tests,” Journal of Finance, vol. 55, no. 3, pp. 1263–1295, 2000. viii [13] R. Litzenberger and A. Kraus, “SKEWNESS PREFERENCE AND THE VALUATION OF RISK ASSETS,” The Journal of Finance, vol. 31, no. 4, pp. 1085 1100, 1976. viii [14] F. Black, “Estimating Expected Returns,” Financial Analysts Journal, vol. 49, no. 5, pp. 36–38, 1993. 1 [15] H. Konno and T. Koshizuka, “Mean-absolute deviation model,” IIE Transactions (Institute of Industrial Engineers), vol. 37, no. 10, pp. 893–900, 2005. 1 [16] F. Black and R. Litterman, “Global Portfolio Optimization,” Financial Analysts Journal, vol. 48, no. 5, pp. 28–43, 1992. 1 [17] V. N. Vapnik, “An overview of statistical learning theory,” IEEE Transactions on Neural Networks, vol. 10, no. 5, pp. 988–999, 1999. 1 [18] D. E. Goldberg, K. Deb, and B. Korb, “Messy Genetic Algorithms Revisited: Nonuniform Size and Scale,” Complex Systems, vol. 4, no. 4, pp. 415–444, 1990. 2 [19] D. E. Goldberg and B. Korb, “An Investigation of Messy Genetic Algorithms,” The Clearinghouse for Genetic Algorithms, The University of Alabama, Tuscaloosa, AL., Tech. Rep., 1990. 2 [20] C. Darwin, On the Origin of Species by Means of Natural Selection Or the Preservation of Favoured Races in the Struggle for Life. H. Milford; Oxford University Press, 1859. 2, 7, 9 [21] T. H. Morgan, “Sex Limited Inheritance in Drosophila,” Science, vol. 32, no. 812, pp. 120–122, 1910. 2 [22] S. Forrest and M. Mitchell, “Relative Building-Block Fitness and the Building- Block Hypothesis,” in In D. Whitley (ed.) Foundations of Genetic Algorithms 2, Morgan Kaufmann, San Mateo, CA., 1993, pp. 109–126. 2 [23] K. J. Oh, T. Y. Kim, and S. Min, “Using genetic algorithm to support portfolio optimization for index fund management,” Expert Systems with Applications, vol. 28, no. 2, pp. 371–379, 2005. 3 [24] A. Gorgulho, R. Neves, and N. Horta, “Applying a GA kernel on optimizing technical analysis rules for stock picking and portfolio composition,” Expert Systems with Applications, vol. 38, no. 11, pp. 14 072–14 085, 2011. 3 [25] P. C. Lin and P. C. Ko, “Portfolio value-at-risk forecasting with GA-based extreme value theory,” Expert Systems with Applications, vol. 36, no. 2 PART 1, pp. 2503–2512, 2009. 3 [26] X. Yang, “Improving portfolio efficiency: A Genetic Algorithm approach,” Computational Economics, vol. 28, no. 1, pp. 1–14, 2006. 3 [27] T. J. Chang, S. C. Yang, and K. J. Chang, “Portfolio optimization problems in different risk measures using genetic algorithm,” Expert Systems with Applications, vol. 36, no. 7, pp. 10 529–10 537, 2009. 3 [28] H. Lee, “Off-Line Building Block Identification in Genetic Algorithms,” Master’s thesis, National Taiwan University, 2012. 8 [29] C. I. Chen, “Development, Management and Investment Strategy of Exchange Traded Fund,” Master’s thesis, National Taiwan University, 2009. 12 [30] P. H. Chen, “An Empirical Study on Relationship between Active Management Strategy and Fund Performance ‘‘ Taiwan Equity Mutual Funds,” Master’s thesis, National Taiwan University, 2012. 13 [31] T. F. Su, “An Investigation of Short- and Long-term Tracking Performance of Leveraged and Inverse ETFs,” Master’s thesis, National Taiwan University, 2015. 15 [32] S. H. Jhao, “Momentum Effect and Economic Cycle in Taiwan,” Master’s thesis, National Taiwan University, 2016. 15 [33] H. W. Chuang, “Revisit The Momentum Strategy- Residual Analysis and Price Risk Adjustment,” Ph.D. dissertation, National Taiwan University, 2011. 16 [34] W. S. Mcculloch and W. Pitts, “A LOGICAL CALCULUS OF THE IDEAS IMMANENT IN NERVOUS ACTIVITY,” BULLETIN OF MATHEMATICAL BIOPHYSICS, vol. 5, 1943. 20 [35] P. J. Werbos, The roots of backpropagation: from ordered derivatives to neural networks and political forecasting. John Wiley Sons, 1994, vol. 1. 21 [36] G. West, “An introduction to Modern Portfolio Theory: Markowitz, CAP-M, APT and Black-Litterman,” Financial Modelling Agency, Tech. Rep., 2006. 22, 23, 51 [37] W. F. Sharpe, “Capital Asset Prices: A Theory of Market Equilibrium under Conditions of Risk,” The Journal of Finance, vol. 19, no. 3, pp. 425–442, 1964. 26 [38] M. Lettau and S. Ludvigson, “Resurrecting the (C)CAPM: A cross-sectional test when risk premia are time-varying,” Journal of Political Economy, vol. 109, no. 6, pp. 1238–1287, 2001. 27 [39] J. Moody and C. J. Darken, “Fast Learning in Networks of Locally-Tuned Processing Units,” Neural Computation, vol. 1, no. 2, pp. 281–294, 1989. 27 [40] B. E. Boser, V. N. Vapnik, and I. M. Guyon, “Training Algorithm Margin for Optimal Classifiers,” Perception, pp. 144–152, 1992. 28 [41] M. P. Kennedy and L. O. Chua, “Neural Networks for Nonlinear Programming,” IEEE Transactions on Circuits and Systems, vol. 35, no. 5, pp. 554–562, 1988. 28 [42] D. S. Broomhead and D. Lowe, “Radial basis functions, multi-variable functional interpolation and adaptive networks,” no. 4148, 1988. 28 [43] J. Lewellen and S. Nagel, “The conditional CAPM does not explain asset pricing anomalies,” Journal of Financial Economics, vol. 82, no. 2, pp. 289–314, 2006. 38 [44] D. E. Goldberg, K. Deb, and J. H. Clark, “Genetic Algorithms, Noise, and the Sizing of Populations,” Department of General Engineering, University of Illinois at Urbana-Champaign, Urbana,IL., Tech. Rep., 1991. 45 [45] T. L. Yu, K. Sastry, D. E. Goldberg, and M. Pelikan, “Population sizing for entropy-based model building in discrete estimation of distribution algorithms,” in Proceedings of GECCO 2007: Genetic and Evolutionary Computation Conference, no. 1, 2007, pp. 601–608. 46 [46] J. F. Bonnans, J. C. Gilbert, C. Lemar´echal, and C. Sagastiz´abal, Optimisation Num´erique: aspects th´eoriques et pratiques. Springer Berlin, Germany, 1997. 51 | |
| dc.identifier.uri | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/56672 | - |
| dc.description.abstract | 本文提出了一種基於基因遺傳演算法(Genetic Algorithm, GA)的最優化風險指標為基礎的均值-方差投資組合模型，這意味著透過設計基因遺傳演算法中的「染色體基因」來表示每個各別資產在投資組合中的權重，近年來在利用基因遺傳演算於優化金融投資組合的領域中，大多都還是傾向以產生傳統的技術指標與均值-方差的隨機分佈組合或僅用於以特定指標作為參考的方向，鮮少人考量到他的市場適用性與可調控性，此文透過調整那些已經過市場與專業基金經理人實證成效的風險指標，例如：詹森指數(Alpha)、潛在上檔報酬比率(Upside-Potential Ratio, UPR)、Risk-on Risk-off指標(RORO)、資產動能(Time Series Momentum, TSM)等，去提供給不同投資偏好的投資者最適當的投資策略，例如:交易時間、波動承受度、預期投資報酬率等，並與傳統的Markowitz 均值-方差效率前緣的投資組合方法與近年比較常見的方法，如使用輻狀基底函數類神經網路(Radial Basis Function Neural Network, RBFN)來優化以均值-方差-偏態為目的的投資組合演算法來相比較，而本文並進而嘗試用人工神經網路(Artificial Neural Network, ANN)去模擬且加速整個基因遺傳演算法的運算過程，經調整後該模型不僅可以依據不同的風險屬性與調整不同的風險指標的偏好程度做為模型的輸入，透過微調輸入的偏好程度與權重之間的關係，實驗結果表明，對於大部分的實驗結果，調整特定參數之後的模型，對於投資者是一種新穎的利用均值-方差來做投資組合問題中權衡輸入條件的一種快速有效的方法，且可以配合不同的風險偏好與投資資產類別來做投資，一方面解決了在 Markowitz投資理論中對於資產輸入太過敏感且過於武斷地考量僅預期收益與歷史波動率，另一方面，解決當輻狀基底函數類神經網路在優化均值-方差-偏態的投資組合時，當偏態的歷史資料過少時，偏態的參考性及有效性將會大幅下降且容易造成過擬合(Overfitting)的情況發生，進而造成最後的結果可能會偏離投資目標的問題，此外，現今在探索財務預測與分類問題中，機器學習的模型在財務預測設計、建構投資組合與風險管理領域都出現許多涉及具有複雜交互關係的大型數據集，目前很難或不可能在完整的經濟模型中充分且有意義地組合使用大量的風險指標，此研究中所提出透過基因遺傳演算法中的多基因組合模型，對於優化不同風險屬性的投資組合問題在未來也提供了更多的使用彈性與調整空間。 | zh_TW |
| dc.description.abstract | In this study, the experiment provides an optimal mean-variance portfolio model with risk indicators based on predictions using a genetic algorithm (GA). It then attempts to simulate and speed up the entire operation process by using neural networks to adjust different risk indicators, such as Jensen’s alpha, upside-potential ratio (UPR), risk-on risk-off (RORO), time-series momentum (TSM). This work finally provides investors who take different investment preferences, such as time, volatility tolerance, expected return on investment with the most appropriate investment strategies. In addition, we compared the approach mentioned above with traditional Markowitz mean-variance efficient frontier of portfolio method, as well as more recently developed methods, such as radial basis function network (RBFN) to simulate the mean-variance-skewness optimize an investment portfolio. In this paper, an artificial neural network is used to simulate this GA process. After adjusting, this model can be used as the input of the model according to different risk preferences or the preference degree of different risk indicators. It can also be based on the different risk preferences or adjust the different indicators of appetite as the input of the model. We try to tune the relationship between preference input and weight finely, but the experimental results show that it is a fast and effective way to use mean and variance to optimize the portfolio with different given conditions by investors. It can be combined with different risk preferences and investment asset classes to invest. On the one hand, the problem solved in Markowitz investment theory for assets input too sensitive and arbitrarily consider, which means only two parameters (historical return and the historical volatility). On the other hand, solving the RBFN mean-variance-skewness problem in the optimal portfolio will encounter the reference, and the effectiveness of the skewness will have fallen sharply to influence the outcome of the last trade-off will deviate from the investment target problem when historical data were few. Moreover, to explore financial projections and machine learning of the classification problems today, the hierarchical model in the financial activity of the design and construction investment portfolio, and risk management has involved much complex data interaction with large data sets. It is hard to combine portfolios to use many risk indicators in the complete economic model meaningfully. This research proposed through multiple genotype combination models of genetic algorithms and provides the unique use of different issues more flexible and adjustable in the future. | en |
| dc.description.provenance | Made available in DSpace on 2021-06-16T05:41:17Z (GMT). No. of bitstreams: 1 U0001-2407202016570000.pdf: 7078340 bytes, checksum: 906d4172b360d4e404926f4314b751e9 (MD5) Previous issue date: 2020 | en |
| dc.description.tableofcontents | 口試委員會審定書 i 誌謝 iii 中文摘要 v Abstract vii List of Figures xiii List of Tables xix 1 Introduction 1 1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.2 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 1.3 Contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 2 Related Work 7 2.1 Genetic Algorithms . . . . . . . . . . . . . . . . . . . . . . . . . 7 2.1.1 Evaluation . . . . . . . . . . . . . . . . . . . . . . . . . 9 2.1.2 Selection . . . . . . . . . . . . . . . . . . . . . . . . . . 9 2.1.3 Crossover . . . . . . . . . . . . . . . . . . . . . . . . . . 10 2.1.4 Mutation . . . . . . . . . . . . . . . . . . . . . . . . . . 10 2.2 Chromosome Design and Population . . . . . . . . . . . . . . . . 11 2.3 Introduction to ETF Markets . . . . . . . . . . . . . . . . . . . . 12 2.3.1 Mean of ETF . . . . . . . . . . . . . . . . . . . . . . . . 13 2.3.2 Type of ETF . . . . . . . . . . . . . . . . . . . . . . . . 13 2.4 Technical Analysis . . . . . . . . . . . . . . . . . . . . . . . . . 15 2.4.1 Filter System . . . . . . . . . . . . . . . . . . . . . . . . 15 2.4.2 Relative Strength Indicators . . . . . . . . . . . . . . . . 15 2.4.3 Price Volume System . . . . . . . . . . . . . . . . . . . . 16 2.5 Sentiment Indicator . . . . . . . . . . . . . . . . . . . . . . . . . 16 2.5.1 Fear Index (VIX) . . . . . . . . . . . . . . . . . . . . . . 16 2.5.2 RORO . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 2.6 Concepts of Neural Networks . . . . . . . . . . . . . . . . . . . . 19 2.6.1 Model of a Neuron . . . . . . . . . . . . . . . . . . . . . 19 2.6.2 The Backpropagation Algorithm . . . . . . . . . . . . . . 21 2.7 Portfolio . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 2.7.1 Markowitz Portfolio Theory . . . . . . . . . . . . . . . . 23 2.7.2 Mean Variance Model . . . . . . . . . . . . . . . . . . . 23 2.7.3 Diversification of Investment . . . . . . . . . . . . . . . . 24 2.8 Capital Asset Pricing Model . . . . . . . . . . . . . . . . . . . . 26 2.8.1 The Single Index Model . . . . . . . . . . . . . . . . . . 26 2.8.2 Beta . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 2.9 RBFN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 3 Methodology 31 3.1 Experimental Goal . . . . . . . . . . . . . . . . . . . . . . . . . 31 3.2 Underlying Assets . . . . . . . . . . . . . . . . . . . . . . . . . . 32 3.2.1 HACK . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 3.2.2 HDV . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 3.2.3 QUAL . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 3.2.4 USMV . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 3.2.5 PUI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 3.2.6 ESG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 3.2.7 MTUM . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 3.2.8 SIZE . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 3.2.9 XLK . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 3.2.10 SOXX . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 3.3 Benchmark . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 3.3.1 SPY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 3.4 Investment Indicators . . . . . . . . . . . . . . . . . . . . . . . . 37 3.4.1 UPR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 3.4.2 Alpha . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 3.4.3 Momentum . . . . . . . . . . . . . . . . . . . . . . . . . 39 3.5 Fitness Function . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 3.6 Chromosome Design . . . . . . . . . . . . . . . . . . . . . . . . 42 3.7 Selection and Crossover . . . . . . . . . . . . . . . . . . . . . . . 44 3.8 RBFN in Finance Portfoilo . . . . . . . . . . . . . . . . . . . . . 46 3.9 Experimental Protocols . . . . . . . . . . . . . . . . . . . . . . . 48 4 Experiment 51 4.1 Efficient Frontier . . . . . . . . . . . . . . . . . . . . . . . . . . 51 4.1.1 20 Days . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 4.1.2 60 Days . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 4.1.3 120 Days . . . . . . . . . . . . . . . . . . . . . . . . . . 54 4.1.4 240 Days . . . . . . . . . . . . . . . . . . . . . . . . . . 55 4.2 Selecting Portfolio Process of Genetic Algorithm . . . . . . . . . 56 4.2.1 20 Days . . . . . . . . . . . . . . . . . . . . . . . . . . . 56 4.2.2 60 Days . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 4.2.3 120 Days . . . . . . . . . . . . . . . . . . . . . . . . . . 61 4.2.4 240 Days . . . . . . . . . . . . . . . . . . . . . . . . . . 63 4.3 Simulate the Process of Genetic Algorithm with Neural Network . 66 4.3.1 Model Structure . . . . . . . . . . . . . . . . . . . . . . 67 4.3.2 Model Training Process . . . . . . . . . . . . . . . . . . 69 4.4 An Interesting Phenomenon of Portfolio Weights Changing . . . . 69 4.4.1 Gain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 4.4.2 Alpha . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72 4.4.3 UPR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74 4.4.4 Absolute Momentum . . . . . . . . . . . . . . . . . . . . 76 4.4.5 Relative Momentum . . . . . . . . . . . . . . . . . . . . 78 4.5 Compare with Different Indicators in NN . . . . . . . . . . . . . 80 4.5.1 15 Days Performance . . . . . . . . . . . . . . . . . . . . 80 4.5.2 30 Days Performance . . . . . . . . . . . . . . . . . . . . 83 4.5.3 60 Days Performance . . . . . . . . . . . . . . . . . . . . 86 4.5.4 120 Days Performance . . . . . . . . . . . . . . . . . . . 89 4.6 Compare with GA, NN, RBFN and EF . . . . . . . . . . . . . . . 92 4.6.1 15 Days Performance . . . . . . . . . . . . . . . . . . . . 92 4.6.2 30 Days Performance . . . . . . . . . . . . . . . . . . . . 95 4.6.3 60 Days Performance . . . . . . . . . . . . . . . . . . . . 98 4.6.4 120 Days Performance . . . . . . . . . . . . . . . . . . . 101 5 Conclusion 105 5.1 Discussion and Future Work . . . . . . . . . . . . . . . . . . . . 105 5.1.1 Portfolio Optimization Process . . . . . . . . . . . . . . . 105 5.1.2 Performance Comparison . . . . . . . . . . . . . . . . . . 106 Reference 109 | |
| dc.language.iso | en | |
| dc.subject | 基因遺傳演算法 | zh_TW |
| dc.subject | 多風險偏好 | zh_TW |
| dc.subject | 風險指標 | zh_TW |
| dc.subject | 投資組合 | zh_TW |
| dc.subject | 最佳化 | zh_TW |
| dc.subject | Genetic Algorithms | en |
| dc.subject | Multi-risk Preference | en |
| dc.subject | Optimizing | en |
| dc.subject | Risk Indicators | en |
| dc.subject | Investment Portfolios | en |
| dc.title | 利用基因遺傳演算法於風險指標最佳化多風險偏好投資組合 | zh_TW |
| dc.title | Optimizing Multi-risk Preference Investment Portfolios by Applying Genetic Algorithms to Risk Indicators | en |
| dc.type | Thesis | |
| dc.date.schoolyear | 108-2 | |
| dc.description.degree | 碩士 | |
| dc.contributor.coadvisor | 蔡芸琤(Yun-Cheng Tsai) | |
| dc.contributor.oralexamcommittee | 于天立(Tian-Li Yu),邱良弼(Ling-Pi Chiu) | |
| dc.subject.keyword | 基因遺傳演算法,投資組合,風險指標,最佳化,多風險偏好, | zh_TW |
| dc.subject.keyword | Genetic Algorithms,Investment Portfolios,Risk Indicators,Optimizing,Multi-risk Preference, | en |
| dc.relation.page | 114 | |
| dc.identifier.doi | 10.6342/NTU202001835 | |
| dc.rights.note | 有償授權 | |
| dc.date.accepted | 2020-07-28 | |
| dc.contributor.author-college | 電機資訊學院 | zh_TW |
| dc.contributor.author-dept | 電機工程學研究所 | zh_TW |
| 顯示於系所單位： | 電機工程學系 | |
文件中的檔案：
| 檔案 | 大小 | 格式 | |
|---|---|---|---|
| U0001-2407202016570000.pdf 未授權公開取用 | 6.91 MB | Adobe PDF |
系統中的文件，除了特別指名其著作權條款之外，均受到著作權保護，並且保留所有的權利。