在线投资组合是近年来计算金融领域热门的研究课题. 目前已有的策略, 对股票价格的预测效果并不十分理想, 而对股价的准确预测对投资组合方式有重要的指导意义. 考虑到股价的滞后性及其分布的复杂性, 首次利用股价中的二阶信息, 提出了DMAR (DMA (Double Moving Average) Reversion)、DEAR (DEA (Double Exponential Average) Reversion)、GMR (GM Reversion)、DA-GMR (DA-GM Reversion) 4种投资组合策略: 分别通过二次移动平均法、二次指数滑动预测法、灰色预测法, 对下一期的价格数据进行了预测、集成学习; 将二次平滑预测和灰色预测的结果进行了优化, 得到了下一期的预测价格; 再利用被动攻击(Passive-Aggressive, PA)算法更新投资组合, 最终得到了4种投资组合策略, 并在真实的金融市场的数据集中验证了策略的有效性. 结果表明, 与已有的算法相比, 在NYSE (O)、NYSE (N)、DJIA和MSCI 这4个真实的金融市场的数据集上, 所提出的4种投资组合策略都达到了较高的累计收益.
In recent years, online portfolios have been a popular topic of research in computational finance. The existing strategies used to forecast stock prices is not ideal, and accurate prediction of stock prices is important for evaluating investment portfolios. Considering the lag in stock prices and the complexity of their distribution, this paper makes use of second-order information in stock prices, for the first time, and proposes four strategies, namely DMAR(DMA(double moving average) reversion), DEAR(DEA(double exponential average) reversion), GMR(GM reversion), and DA-GMR(DA-GM reversion). The second-order moving average method, the second exponential sliding prediction method, and the gray prediction method are used to predict price data for the next period; integrated learning optimizes the results of second-order smoothing prediction and the gray prediction is used to obtain the predicted price. Next, we use PA(passive-aggressive) algorithms to update the portfolio, and we arrive at four portfolio strategies. We verify the effectiveness of these strategies using real data from the financial market. The results show that compared with existing algorithms, the four strategies proposed in this paper achieved higher cumulative returns on datasets for NYSE(O), NYSE(N), DJIA, and MSCI.
[1] STEINBACH M C. Markowitz revisited: Mean-variance models in financial portfolio analysis [J]. SIAM Review, 2001, 43(1): 31-85
[2] KELLYJ L. A new interpretation of information rate [J]. IRE Transactions on Information Theory, 1956, 2(3): 185-189
[3] COVER M, ORDENTLICH E. Universal portfolios with side information [J]. IEEE Transactions on Information Theory, 1996, 42(2): 348-363
[4] SINGER Y. Switching portfolios [J]. International Journal of Neural Systems, 1997, 8(4): 445-455
[5] HELMBOLD D P, SCHAPIRE R E, SINGER Y, et al. On-line portfolio selection using multipli-cative updates [J]. Mathematical Finance, 1998, 8(4): 243-251
[6] GAIVORONSKI A A, STELLA F. Stochastic nonstationary optimization for finding universal portfolios [J]. Annals of Operations Research, 2000, 100: 165-188
[7] FAGIUOLI E, STELLA F, VENTURA A. Constant rebalanced portfolios and side-information [J]. Quantitative Finance, 2007, 7(2): 161-173
[8] AGARWAL A, HAZAN E, KALE S, et al. Algorithms for portfolio management based on the Newton method [C]//Proceedings of the 23rd International Conference on Machine learning. ACM, 2006: 9-16.
[9] BORODIN A, EL-YANIV R, GOGAN V. Can we learn to beat the best stock [J]. Journal of Artificial Intelligence Research, 2004, 21: 579-594
[10] LI B, ZHAO P L, HOI S C H, et al. PAMR: Passive aggressive mean reversion strategy for portfolio selection [J]. Machine Learning, 2012, 87(2): 221-258
[11] LI B, HOI S C H, ZHAO P L, et al. Confidence weighted mean reversion strategy for online portfolio selection [J]. ACM Transactions on Knowledge Discovery from Data, 2011, 15(1): 434-442
[12] LI B, HOI S C H. On-line portfolio selection with moving average reversion [C]// Proceedings of the 29th International Conference on Machine Learning. [S.l]: [s.n.], 2012: 563-570.
[13] HUANG D J, ZHOU J L, LI B, et al. Robust median reversion strategy for online portfolio selection [J]. IEEE Transactions on Knowledge and Data Engineering, 2016, 28(9): 2480-2493
[14] LI B, WANG J L, HUANG D J, et al. Transaction cost optimization for online portfolio Selection [J]. Quantitative Finance, 2017, 18(8): 1411-1424
[15] GYÖRFI L, UDINA F, WALK H. Nonparametric nearest neighbor based empirical portfolio selection strategies [J]. Statistics and Decisions, 2008, 26(2): 145-157
[16] GYÖRFI L, LUGOSI G, UDINA F. Nonparametric kernel-based sequential investment strategies [J]. Mathematical Finance, 2006, 16(2): 337-357
[17] LI B, HOI S C H, GOPALKRISHNAN V. CORN: Correlation-driven nonparametric learning approach for portfolio selection [J]. ACM Transactions on Intelligent Systems and Technology, 2011, 2(3): Article number 21
[18] DUFFY N, HELMBOLD D. Boosting methods for regression [J]. Machine Learning, 2002, 47(2/3): 153-200
[19] CLAESKENS G, MAGNUS J R, VASNEV A L, et al. The forecast combination puzzle: A simple theoretical explanation [J]. International Journal of Forecasting, 2016, 32(3): 754-762
[20] CRAMMER K, DEKEL O, KESHET J, et al. Online passive-aggressive algorithms [J]. Journal of Machine Learning Research, 2006, 7(3): 551-585
[21] POTERBA J M, SUMMERS L H. Mean reversion in stock prices: Evidence and implications [J]. Social Science Electronic Publishing, 1987, 22(1): 27-59
[22] GRINOLD R C, KAHN R N. Active portfolio management: A quantitative approach for providing superior returns and controlling risk [M]. 2nd ed. New York: McGraw-Hill, 1999
中国综合性科技类核心期刊(北大核心)