基于非线性方法和VaR的均线交易系统研究
|
摘要
世界范围内的金融市场正处于一个迅速发展的历史时期，近年来，交易系统的研究与应用在国外得到了快速的发展，在交易中占有越来越大的比重。根据纽约证券交易所公布的数据，截至2011年5月20日当周该交易所的日均程序化交易占比为28.6%。
成功的交易系统能够产生稳定和超额的回报，根据美国权威交易系统评选杂志2011年发布的交易系统排名，前三名模型年收益率均在200%以上。
交易系统通常包含有追求最大收益率的阿尔法模型和以控制风险敞口规模为主的风险控制模型。但当前国内外对交易系统的研究主要倾向于交易信号的设计和挖掘，试图以一致的方法在任何的趋势中获利,没有注意到交易系统本身所存在的缺陷。在对交易系统的风险控制模型的研究较少，将阿尔法模型和风险控制模型两者结合在一起的研究尚不多见。
为解决上述研究存在的不足，本文以传统技术分析中的均线交易系统为基础，使用支持向量机（SVM）、多目标优化算法中的非支配解和风险管理的VaR方法，构建了交易系统中重要的两个模型：阿尔法模型和风险控制模型，形成了基于非线性方法和VaR的均线交易系统。
传统的均线交易系统在趋势市场中具有明显的赢利效应，但在横盘市场中却反复亏损。针对交易系统的这个缺陷，本文首先利用SVM分类器对市场进行趋势识别，使用RAVI等5种趋向技术指标将股票价格时间序列映射到高维特征空间，构建了支持向量机分类器对趋势进行分类和过滤，对不利于均线系统交易的横盘趋势进行过滤（空仓），以上证指数为研究对象，将5-60日均线作为基本参数，改进基于趋势跟随的均线交易系统，建立了基于SVM分类器的均线交易系统。
在这个基础上，进一步优化参数。在参数优化过程中，为防止出现参数的过度拟合，将交易系统中常用且重要的两个评价指标，最大收益与连续最大回撤作为目标，使用了多目标优化算法中的非支配解的方法。经过优化，完成了对交易模型中的一个重要的组成部分-阿尔法模型的构建。
为建立风险控制模型，本文以5-60日均线交易系统为研究对象，建立了非特定时间动态VaR模型。用蒙特卡罗方法产生了近3000个交易收益率数据、分析了非特定时间动态VaR收益率分布特征，并进行了模型准确性检验；在使用非特定时间动态VaR模型进行风险管理后，研究结果表明可以优化交易策略。因此研究完成了对非特定时间动态VaR模型-风险控制模型的构建。
最后将阿尔法模型与风险控制模型组合起来，构建了基于非线性方法和VaR的均线交易系统。为了将非特定时间动态VaR模型引入，首先使用威尔科克森秩和检验的方法验证了使用SVM前后，交易系统所生成的收益率序列的VaR值在置信条件下是没有统计差别的。然后通过对参数优化后的均线交易系统进行动态VaR建模求解。结果表明，基于非线性方法和VaR的均线交易系统可以有效地提高收益和降低风险。
将非线性方法和VaR方法与投资交易相结合，有利于推动非线性科学在投资领域的应用，同时基于非线性方法和VaR的均线交易系统的构建也为投资者提供了一整套科学的投资方法，丰富了投资的研究方法，为程序化交易在中国股市的应用提供经验证据。
Chinese financial market is in the midst of a rapid developing period while global financial market is also experiencing profound changes.As the traditional theory is unable to keep up with the rapid development of financial market, the program trading system is playing a much more important part in the market, and the trading system has already been appliedwidely in Europe and America by now. According to the reporting data published by NYSE, the average amount of dailytrading exchangingviaprogram tradinghas reached28.6%of the weekby20th May2011. A successfully operatingtrading system would generate stable and excessive returns in stock market. According to the trading system ranking of2011released by a respected trading system evaluation magazine, the annual rate of return of the top three models were above200%.
Currently, the research hot points of domestic and overseasstill focus on trend prediction in trading system area, but the issues of prediction accuracy remain unresolved big challenge.Thoughthe researches mainly attempt to make profit in any trend by using consistent method, the defects of the trading system are not been paidenough attention to. As a result, the risk control of the trading system lacks a scientific and effective method and no researches with VaRmodels and tools are applied in this field.
To make up for the deficiency of the above researches and based on the simple moving average trading method of traditional technical analysis, support vector machine (SVM), the non-dominated solution of multi-objective optimization algorithm and VaR method of risk management are applied to develop two important models in the trading system including alpha model and risk control model and constitute the moving average trading system based on nonlinear method and VaR.
The Alpha refers to the investment return deducing the market benchmark return. Alpha model represents a trading system used forselectinga trading timing during the investment to make profit. Risk management is not only used to avoid the risk or minimize losses, but also to implement purposeful selection and scale control on risk exposures thus to improve the quality and continuity of return. Although the profit would be influenced by risk control model, more robust benefit is brought out by reducing return volatility.
The simple moving average trading system has obvious profitable effect in trend market, but would suffer from repeated losses in sideways market. In order to filter the unfavorable trends, SVM classifier is used to identify market trends, and5tendency technical indicators including RAVI are used to map the share prices time series to a high-dimensional feature space.Then, support vector machine classifier is constructed to classify and filter trends through which fluctuation trend and downward trend are filtered (short position). Taking SSE composite index as research object and5-60days moving average as the basic rules, the simple moving average system based on trend following is improved and SVM classifier is established.
On this basis, non-dominated solution approach is used to optimize the parameters of the simple moving average system based on SVM classifier. During parameters optimization process, two common and important evaluation indicators in the trading system, maximum return and continuous maximum drawdown, are taken as objectives in order to prevent the over fitting of parameters. Therefore, the non-dominated solution approach in multi-objective optimization algorithm is used for further optimizing parameters. The construction of an important integral part of the trading model–alpha model–is accomplished through optimization.
With the purpose of establishing a risk control model, nonspecific time dynamic VaR model specific to the simple moving average trading system is built up. Taking5-60days moving average as research object, the nonspecific time dynamic VaR model is built and nearly3000trading rate of return data are generated and the distribution characteristics of the nonspecific time dynamic VaR rate of turn are analyzed by Monte Carlo method.Furthermore, the accuracy tests are conducted on the models respectively.Risk management via nonspecific time dynamic VaR model would optimize the trading strategies.By building nonspecific time dynamic VaR model and testing the accuracy of the model, VaR model is well applied in nonspecific time dimension and makes significant application value in the risk control for the trading system.
Finally, by combining optimized moving average trading model based on SVM which is the alpha model and the nonspecific time dynamic VaR model which is risk control model, a simple moving average trading system based on nonlinear method and VaR is formed. To imply nonspecific time dynamic VaR model, Wilcoxon's rank sum test is used to compare the results before and after SVM and it is found that there is no statistical difference between the VaR values of return series generated by the trading system under the condition of confidence. After that, the moving average system with parameter optimization is modeled by dynamic VaR and solved.The results show that it is efficient to apply nonlinear method and VaR into the simple moving average trading model in order to make compound return and risk control. Meanwhile, the return profit is improved and risk is reduced.
This trading system proposed above is contributed to the establishing process of Chinese financial market, especially to improve the efficiency of resources allocation. It is also a certain new concept and of realistic value. From a microscopic view, the results is helpful to the investors to behave more rationally in the trading market. On the other hand,it facilitates the perfection of market regulation and the resource allocation optimization from the macroscopic view.
Combining the nonlinear methodologies withVaR method and applied the new model into the investment trading system is helpful top romote the application of nonlinear science in the investment field.Meanwhile, the building of thesimple moving average trading model based on nonlinear method and VaR also provides investors with a complete set of scientific investment method and enriching investment research techniques,which is alsoaccumulating experiences in applying program trading in Chinese stock market.
成功的交易系统能够产生稳定和超额的回报，根据美国权威交易系统评选杂志2011年发布的交易系统排名，前三名模型年收益率均在200%以上。
交易系统通常包含有追求最大收益率的阿尔法模型和以控制风险敞口规模为主的风险控制模型。但当前国内外对交易系统的研究主要倾向于交易信号的设计和挖掘，试图以一致的方法在任何的趋势中获利,没有注意到交易系统本身所存在的缺陷。在对交易系统的风险控制模型的研究较少，将阿尔法模型和风险控制模型两者结合在一起的研究尚不多见。
为解决上述研究存在的不足，本文以传统技术分析中的均线交易系统为基础，使用支持向量机（SVM）、多目标优化算法中的非支配解和风险管理的VaR方法，构建了交易系统中重要的两个模型：阿尔法模型和风险控制模型，形成了基于非线性方法和VaR的均线交易系统。
传统的均线交易系统在趋势市场中具有明显的赢利效应，但在横盘市场中却反复亏损。针对交易系统的这个缺陷，本文首先利用SVM分类器对市场进行趋势识别，使用RAVI等5种趋向技术指标将股票价格时间序列映射到高维特征空间，构建了支持向量机分类器对趋势进行分类和过滤，对不利于均线系统交易的横盘趋势进行过滤（空仓），以上证指数为研究对象，将5-60日均线作为基本参数，改进基于趋势跟随的均线交易系统，建立了基于SVM分类器的均线交易系统。
在这个基础上，进一步优化参数。在参数优化过程中，为防止出现参数的过度拟合，将交易系统中常用且重要的两个评价指标，最大收益与连续最大回撤作为目标，使用了多目标优化算法中的非支配解的方法。经过优化，完成了对交易模型中的一个重要的组成部分-阿尔法模型的构建。
为建立风险控制模型，本文以5-60日均线交易系统为研究对象，建立了非特定时间动态VaR模型。用蒙特卡罗方法产生了近3000个交易收益率数据、分析了非特定时间动态VaR收益率分布特征，并进行了模型准确性检验；在使用非特定时间动态VaR模型进行风险管理后，研究结果表明可以优化交易策略。因此研究完成了对非特定时间动态VaR模型-风险控制模型的构建。
最后将阿尔法模型与风险控制模型组合起来，构建了基于非线性方法和VaR的均线交易系统。为了将非特定时间动态VaR模型引入，首先使用威尔科克森秩和检验的方法验证了使用SVM前后，交易系统所生成的收益率序列的VaR值在置信条件下是没有统计差别的。然后通过对参数优化后的均线交易系统进行动态VaR建模求解。结果表明，基于非线性方法和VaR的均线交易系统可以有效地提高收益和降低风险。
将非线性方法和VaR方法与投资交易相结合，有利于推动非线性科学在投资领域的应用，同时基于非线性方法和VaR的均线交易系统的构建也为投资者提供了一整套科学的投资方法，丰富了投资的研究方法，为程序化交易在中国股市的应用提供经验证据。
Chinese financial market is in the midst of a rapid developing period while global financial market is also experiencing profound changes.As the traditional theory is unable to keep up with the rapid development of financial market, the program trading system is playing a much more important part in the market, and the trading system has already been appliedwidely in Europe and America by now. According to the reporting data published by NYSE, the average amount of dailytrading exchangingviaprogram tradinghas reached28.6%of the weekby20th May2011. A successfully operatingtrading system would generate stable and excessive returns in stock market. According to the trading system ranking of2011released by a respected trading system evaluation magazine, the annual rate of return of the top three models were above200%.
Currently, the research hot points of domestic and overseasstill focus on trend prediction in trading system area, but the issues of prediction accuracy remain unresolved big challenge.Thoughthe researches mainly attempt to make profit in any trend by using consistent method, the defects of the trading system are not been paidenough attention to. As a result, the risk control of the trading system lacks a scientific and effective method and no researches with VaRmodels and tools are applied in this field.
To make up for the deficiency of the above researches and based on the simple moving average trading method of traditional technical analysis, support vector machine (SVM), the non-dominated solution of multi-objective optimization algorithm and VaR method of risk management are applied to develop two important models in the trading system including alpha model and risk control model and constitute the moving average trading system based on nonlinear method and VaR.
The Alpha refers to the investment return deducing the market benchmark return. Alpha model represents a trading system used forselectinga trading timing during the investment to make profit. Risk management is not only used to avoid the risk or minimize losses, but also to implement purposeful selection and scale control on risk exposures thus to improve the quality and continuity of return. Although the profit would be influenced by risk control model, more robust benefit is brought out by reducing return volatility.
The simple moving average trading system has obvious profitable effect in trend market, but would suffer from repeated losses in sideways market. In order to filter the unfavorable trends, SVM classifier is used to identify market trends, and5tendency technical indicators including RAVI are used to map the share prices time series to a high-dimensional feature space.Then, support vector machine classifier is constructed to classify and filter trends through which fluctuation trend and downward trend are filtered (short position). Taking SSE composite index as research object and5-60days moving average as the basic rules, the simple moving average system based on trend following is improved and SVM classifier is established.
On this basis, non-dominated solution approach is used to optimize the parameters of the simple moving average system based on SVM classifier. During parameters optimization process, two common and important evaluation indicators in the trading system, maximum return and continuous maximum drawdown, are taken as objectives in order to prevent the over fitting of parameters. Therefore, the non-dominated solution approach in multi-objective optimization algorithm is used for further optimizing parameters. The construction of an important integral part of the trading model–alpha model–is accomplished through optimization.
With the purpose of establishing a risk control model, nonspecific time dynamic VaR model specific to the simple moving average trading system is built up. Taking5-60days moving average as research object, the nonspecific time dynamic VaR model is built and nearly3000trading rate of return data are generated and the distribution characteristics of the nonspecific time dynamic VaR rate of turn are analyzed by Monte Carlo method.Furthermore, the accuracy tests are conducted on the models respectively.Risk management via nonspecific time dynamic VaR model would optimize the trading strategies.By building nonspecific time dynamic VaR model and testing the accuracy of the model, VaR model is well applied in nonspecific time dimension and makes significant application value in the risk control for the trading system.
Finally, by combining optimized moving average trading model based on SVM which is the alpha model and the nonspecific time dynamic VaR model which is risk control model, a simple moving average trading system based on nonlinear method and VaR is formed. To imply nonspecific time dynamic VaR model, Wilcoxon's rank sum test is used to compare the results before and after SVM and it is found that there is no statistical difference between the VaR values of return series generated by the trading system under the condition of confidence. After that, the moving average system with parameter optimization is modeled by dynamic VaR and solved.The results show that it is efficient to apply nonlinear method and VaR into the simple moving average trading model in order to make compound return and risk control. Meanwhile, the return profit is improved and risk is reduced.
This trading system proposed above is contributed to the establishing process of Chinese financial market, especially to improve the efficiency of resources allocation. It is also a certain new concept and of realistic value. From a microscopic view, the results is helpful to the investors to behave more rationally in the trading market. On the other hand,it facilitates the perfection of market regulation and the resource allocation optimization from the macroscopic view.
Combining the nonlinear methodologies withVaR method and applied the new model into the investment trading system is helpful top romote the application of nonlinear science in the investment field.Meanwhile, the building of thesimple moving average trading model based on nonlinear method and VaR also provides investors with a complete set of scientific investment method and enriching investment research techniques,which is alsoaccumulating experiences in applying program trading in Chinese stock market.
引文
[1]徐智斌.行为金融学对经典金融学的修正与创新[J].商业时代,2007,(4):22-25.
[2]埃德加. E.彼得斯.资本市场的混沌与秩序[M].经济科学出版社,1997.
[3]马丁.技术分析[M].中国财政经济出版社,2003:183-390.
[4]李莎,李红刚.股票市场中技术分析有效性的实证研究[J].北京师范大学学报,(自然科学版),2009,(4):212-214.
[5]魏玉根.技术交易系统与我国股市有效性的实证分析[J].经济科学,2000,(2):56-63.
[6]张维,赵帅特,熊熊,张永杰.计算实验金融,技术规则与时间序列收益可预测性[J].管理科学,2008,(6):74-84.
[7] Fernandez-Rodríguez, González-Martel, Sosvilla-Rivero. Optimization ofTechnical Rules by Genetic Algorithms: Evidence from the Madrid StockMarket[J]. Applied Financial Economics,2005,(15):773-775.
[8] Akbar Esfahanipour, Somayeh Mousavi. A Genetic Programming Model toGenerate Risk-adjusted Technical Trading Rules in Stock Market[J]. ExpertSystems with Applications,2011,38:8438-8445.
[9] Creamer, Freund. Automated Trading with Boosting and ExpertWeighting[J]. Quantitative Finance,2010,(10):401-420.
[10] Wilson, Banzhaf. Algorithmic Trading With Developmental And LinearGenetic Programming[J]. Genetic Programming Theory And Practice Vii,2010:119-134.
[11] Katrina Clay. Program Trading Averaged28.6Percent of NYSE Volumeduring May16-20[M/OL].2011,(6).\\http://www. nyse. com/press/1302173178545. html.
[12]杨喆.全球顶尖程序化交易模型研究[R].量化研究专题报告,海通证券,2012,(2).
[13]闫冀楠,张维,刘志弘.金融技术发展的智能化趋势-人工智能及其杂合对当代金融理论与实践的影响[J].河南金融管理干部学院学报,1998,(5):14-16.
[14] Anis A. A., Llogy E. H. The Expected Value of the Adjusted RescaledHurst Range of Independent National Summands. Biometrical Journal.Black F, Scholes M. The Pricing of Options and Corporate Liabilities. TheJournal of Political Economy,1996,(3).
[15] Daniela Federici, Giancarlo Gandlfo. Chaos and The Exchange Rate.Journal oF International Trade and Economic Development,2002,(5).
[16] Hai Chin Yu, Ming Chang Huang. Statistical Properties of Volatility inFractal Dimension and Probability Distribution among Six Stock Markets.Applied Financial Economics,2004,(7).
[17] Klaasjan Visscher,, Arie Rip. Coping with Chaos in Change Processes.Creativity and Innovation Management,2003.
[18] Richard H,Day, Oleg V. Pavlov. Computing Economic Chaos[J].Computational Economics,2004,(4).
[19]彭晓东,莫东松,刘勇,夏新初,夏克俭.基于数据仓库的综合决策支持系统的设计研究[J].计算机工程与设计,2003,(5):33-36.
[20]冉宇瑶.证券数据仓库在证券业的应用及其结构[J].计算机工程与设计,2004,(8):45-48.
[21]温小敏,顾锋娟.马尔可夫交易策略研究[J].统计与决策,2007,(1).
[22]梁元星.测股市分析股价的随机过程模型的建构[J].广西民族学院学报,(自然科学版),2003,(3).
[23] Brock, Lakonishok, LeBaron. Simple Technical Trading Rules and TheStochastic Properties of Stock Returns[J]. The Journal of Finance XLVII,1992,(5):1731-1764.
[24]周明磊.非参数估计与小波分析在股市趋势线中的应用[J].数理统计与管理,2005,(4).
[25]高雷,任慧玉.基于小波分析的我国股票市场趋势探测[J].统计与决策,2005,(16).
[26] Chan, Jegadeesh, Lakonishok. Momentum Strategies[J]. The Journal ofFinance,1996,(5):1681-1713.
[27] Hill T, Connor M, Remus, W. Neural Network Models for Time SeriesForecasts[J]. Management Science,1996.
[28] Jun Yu. Forecasting Volatility in The New Zealand Stock Market[J]. Journalof Fianancial Economics,2002.
[29] Tao Wang. Based on the Theory of Fuzzy of Stock Evaluation System[J],2008.
[30] NaiYang Deng, YingJie Tie. The New Method on Data Mining—theSupport Vector Machine[J],2004.
[31] NaiYang Deng, YingJie Tie. Support Vector Machine-theory[J]. Algorithmand the Development,2009.
[32] Bo Lu. The Application of Support Vector Machine in the Classification[J].International Journal of Image and Graphics,2005.
[33] Wei Sheng. Model Research of Stock Price Prediction[J]. EconomicProblems Research,2009.
[34] Neftci., Nanve Trading Rules in Financial Markets and Wiener KolmogorovPrediction Theory: A Study of Technical Analysis[J]. Journal of Business,1991,(64):549-571.
[35] Nez-Letamendia. Fitting the Control Parameters of a Genetic Algorithm: anApplication to Technical Trading Systems Design[J]. European Journal ofOperational Research,2007,(179):847-868.
[36] Andrew. Szakmary, Qian Shen, Subhash. Trend-following TradingStrategies In Commodity Futures: A re-examination[J]. Journal of Banking&Finance,2010,(34):409-426.
[37]王庆宗.移动平均线建构动量策略研究[J].河南科技,2010,(14):21-23.
[38]苏匡,罗元,方楚贤.基于神经网络对我国股票市场技术分析的实证检验[J].当代经济,2010,(7):146-147.
[39]尹宇明.中国股市技术交易规则有效性的计量分析[J].现代财经(天津财经大学学报),2011,(6):81-87.
[40]韦增欣,周智超,韦鑫.技术分析在A股市场的有效性研究及实证检验[J].重庆理工大学学报(自然科学),2013,(7):113-120.
[41]许红伟,吴冲锋,张翔.我国量化（对冲）基金业绩表现的比较分析[J].投资研究,2013,(2):72-87.
[42] Dourra, Siy. Investment using Technical Analysis and Fuzzy Logic[J].Fuzzy sets and Systems,2002,(127):221-240.
[43] George S. Atsalakis, Kimon P. Valavanis. Surveying Stock MarketForecasting Techniques–Part II: Soft Computing Methods[J]. ExpertSystems with Applications,2009,(36):5932-5941.
[44] L. Dymova, P. Sevastianov, P. Bartosiewicz.A New Approach to TheRule-base Evidential Reasoning: Stock Trading Expert System Application[J].Expert Systems with Applications,2010,(37):5564-5576.
[45] Pei-Chann Chang, Chin-Yuan Fan, Jun-Lin Lin.Trend Discovery inFinancial Time Series Data using a Case Based fuzzy Decision Tree[J].Expert Systems with Applications,2011,(38):6070-6080.
[46]张杨,宋恒.基于聚类技术的股市基本趋势规律挖掘[J].数理统计与管理,2006,(7):18-21.
[47]汪东.基于支持向量机的选股与选时研究[J].博士毕业论文,上海交通大学,2007.
[48]赵尔波,马欢,韩战钢.遗传程序设计分析股价移动平均及中长期走势[J].计算机应用研究,2010,(6):17-18.
[49]谢丁.反趋向类技术分析指标的实证研究及评测[J].黑龙江金融,2011,(3):38-40.
[50]苏治,傅晓媛.核主成分遗传算法与SVR选股模型改进[J].统计研究,2013,(5):54-62.
[51] Pruitt, White R. The CRISMA Trading System: Who Says TechnicalAnalysis Can’t Beat the Market?[J]. Journal of Portfolio Management,1988,(1):55-58.
[52] Potvin, Soriano, Vallte. Generating Trading Rules on the Stock Marketswith Genetic Programming[J]. Computers&Operations Research,2004,(31):1033-1047.
[53] Kwon, Moon. A Hybrid Neurogenetic Approach for Stock Forecasting[J].IEEE Transactions on Neural Networks,2007,18(3):851-864.
[54] Abeyratna, David. The Profitability of Moving Average Trading[J]. Rulesin South Asian Stock Markets. Emerging Markets Review,2001:17-33.
[55] Laura Nu′n ez-Letamendia.Fitting the Control Parameters of a GeneticAlgorithm: An Application to Technical Trading Systems Design[J].European Journal of Operational Research,2007,(179):847-868.
[56] Casqueiro, Rodrigues L. Neuro-dynamic Trading Methods[J]. EuropeanJournal of Operational Research,2006,(175):1400-1412.
[57] Lai Ming-Ming. The Profitability of the Simple Moving Averages andTrading Range Breakout in the Asian Stock Market[J]. Journal of AsianEconomics,2006,(17):144-170.
[58] Stephanos Papadamou, George Stephanides. Improving Technical TradingSystems by Using a New MATLAB-based Genetic Algorithm Procedure[J].Mathematical and computer Modelling,2007,(46):189-197.
[59] Andrada-ftlix, Fernandez-Rodríguez. Improving Moving Average TradingRules with Boosting and Statistical Learning Methods[J]. Journal ofForecasting,2008,(27):433-449.
[60] Thira Chavar nakul, David Enke. Intelligent Technical Analysis Based EquiVolume Charting for Stock Trading using Neural Networks[J]. ExpertSystems with Applications,2008,(34):1004-1017.
[61] Pei-Chann Changa, T. Warren Liaob, Jyun-Jie Lina, Chin-Yuan Fanc.ADynamic Threshold Decision System for Stock Trading Signal Detection[J].Applied Soft Computing,2011,(11):3998-4010.
[62] Massimiliano. Investment using Evolutionary Learning Methods andTechnical Rules[J]. European Journal of Operational Research207,2010:1717-1727.
[63] Hoi Tin Kong, Qing Zhang, G. George Yin. A Trend-following Strategy:Conditions for Optimality[J]. Automatica,2011,(47):661-667.
[64]王兆军,曾渊沧,郝刚.移动平均线方法的最佳步长组合的确定[J].高校应用数学学报A,2000,(2):219-228.
[65]曾宪聚,代文强,冯建栋.基于移动平均线的证券投资决策方法研究[J].经济师,2003,(10):134-136.
[66]刘厉兵.基于马尔可夫链的移动平均线分析新思维[J].商业研究,2005,(12):137-138.
[67]丁圣,高风.小波神经网络在股票平均线交易规则中的应用[J].计算机仿真,2006,(11):259-262.
[68]简清明,曾黄麟.基于移动平均线组合和支持向量机的股市趋势研究[J].计算机应用与软件,2011,(12):81-83.
[69]张伟,李泓仪,兰书梅,张洁. GA-SVM对上证综指走势的预测研究[J].东北师大学报（自然科学版）,2012,(3):55-59.
[70] Allen, Karjalainen. Using Genetic Algorithms to Find Technical TradingRules[J]. Journal of Financial Economics,1999,(51):245-271.
[71] Stephan Schulmeister.Profitability of technical Stock Trading: HAS itMoved From daily to Intraday Data?[J]. Review of Financial Economics,2009,(18):190-201.
[72] Lamartine Almeida Teixeira, Adriano Lorena Inácio de Oliveira.A methodfor Automatic Stock Trading Combining Technical Analysis and NearestNeighbor Classification[J]. Expert Systems with Applications,2010,(37):6885-6890.
[73] Simon,Yan. Trend Following Algorithms In Automated Derivatives MarketTrading[J]. Expert Systems with Applications,2012,(4):7758-7771.
[74]苏涛.金融市场风险VaR度量方法的改进研究.天津大学博士论文,2007.
[75]郑文通.金融风险管理的VAR方法及其应用[J].国际金融研究,1997,(9):76-79.
[76]杨春鹏.金融衍生工具风险管理的概率方法[J].数量经济技术经济研究,1998,(11):36-39.
[77]王春峰,蒋祥林,李刚.基于随机波动性模型的中国股市波动性估计[J].管理科学学报,2003,(4):8-12.
[78]马超群,李红权,徐山鹰,杨晓光,李晖.风险价值的完全参数方法及其在金融市场风险管理中的应用[J].系统工程理论与实践,2001,(4):44-48.
[79]江红莉,何建敏,胡小平.基于时变Copula的La-VaR测度研究[J].重庆大学学报,(社会科学版),2013,(3):27-33.
[80]庄新路,庄新田,黄小原.基于VAR风险指标的投资组合模糊优化[J].数学的实践与认识,2003,(3):55-57.
[81]陈立新. VaR风险测量模型在我国股票市场中的应用[J].大连铁道学院学报,2004,(2):44-48.
[82]蒋虹,曲丹丹.基于VaR的沪深300股指期货风险管理实证研究[J].经济问题,2008,(12):74-80.
[83]张鹏.不允许卖空情况下均值-方差和均值-VaR投资组合比较研究[J].中国管理科学,2008,(4):64-69.
[84]吴中冉,徐强.基于VaR的股指期货套期保值研究[J].金融经济,2010,(8):61-64.
[85]钟纯,吴雪,陈文财.基于VaR和CVaR下优化模型及其在保险资金组合投资中的应用[J].南昌大学学报,(工科版),2011,(4):398-403.
[86]魏波.多因素证券投资组合在VaR风险约束下的模型及实证研究[J].长春工业大学学报,(社会科学版),2013,(4):29-32.
[87]范K撤普.通向金融王国之路[M].机械工业出版社,1998:233-246.
[88]刘伟,杨廷干.关于股市技术分析的风险测度研究[J].金融与经济,2010,(9):57-59.
[89] Tushar S. Chande.超越技术分析[M].万卷出版公司,2010:20-25.
[90]丁鹏.量化投资[M].电子工业出版社,2012:20-23.
[91] Philip Treleaven, Michal Galas, Vidhi Lalc hand. Algorithmic Trading Review[J]. Communications of The Acm,2013,(9):76-85.
[92]奥尔德里奇.打开量化投资的黑箱[M].机械工业出版社,2013:57-61.
[93]张学工.关于统计学习理论与支持向量机[J].自动化学报,(1):32-42.
[94] HsuC, LinC. A Comparison of Methods for Multiclass Support VectorMachines [J]. IEEE Transon Neural Networks,2002,13(2):415-425.
[95] Takahash i F, Abe S. Decision-Tree-Based Multiclass Support VectorMachines[A]. Procof the9th Int Confon Neural Information P rocessing[C].Singapore,2002,(3):1418-1422.
[96] Sungmoon C, Sang H O, Soo-Young L. Support Vector Machines withBinary Tree Arch Itecture for Multi-Class Classification[J]. NeuralInformation Processing-letters and Reviews,2004,2(3):47-51.
[97]唐发明,王仲东,陈绵云.支持向量机多类分类算法研究[J].控制与决策,2005,(7):746-754.
[98][98]Yao X, Liu Y. Search Methodologies. Springer[J]. Chapter MachineLearning,2005:341-373.
[99] T.Amahroq, A. Taa. On Lagrange Kuhn—Tucker Multipliers for MultiObjective Optimization Problems[J]. Optimization,1997,41:159-172.
[100]C oello, Lamont, Veldhuizen. Evolutionary Algorithms for Solving Multi-Objective Problem[J]. Springer,2007.
[101]菲利普,乔瑞.风险价值VaR—金融风险管理新标准[M].北京:中信出版社,2005:251-277.
[102]R uey S. Tsay.金融时间序列分析[M].人民邮电出版社,2009:71-127.
[103]F ishman, G. Monte Carlo: ConCept, Algorithms, and Applieations[J].Springer Series in Operational Researeh. Springer, Berlin,1997.
[2]埃德加. E.彼得斯.资本市场的混沌与秩序[M].经济科学出版社,1997.
[3]马丁.技术分析[M].中国财政经济出版社,2003:183-390.
[4]李莎,李红刚.股票市场中技术分析有效性的实证研究[J].北京师范大学学报,(自然科学版),2009,(4):212-214.
[5]魏玉根.技术交易系统与我国股市有效性的实证分析[J].经济科学,2000,(2):56-63.
[6]张维,赵帅特,熊熊,张永杰.计算实验金融,技术规则与时间序列收益可预测性[J].管理科学,2008,(6):74-84.
[7] Fernandez-Rodríguez, González-Martel, Sosvilla-Rivero. Optimization ofTechnical Rules by Genetic Algorithms: Evidence from the Madrid StockMarket[J]. Applied Financial Economics,2005,(15):773-775.
[8] Akbar Esfahanipour, Somayeh Mousavi. A Genetic Programming Model toGenerate Risk-adjusted Technical Trading Rules in Stock Market[J]. ExpertSystems with Applications,2011,38:8438-8445.
[9] Creamer, Freund. Automated Trading with Boosting and ExpertWeighting[J]. Quantitative Finance,2010,(10):401-420.
[10] Wilson, Banzhaf. Algorithmic Trading With Developmental And LinearGenetic Programming[J]. Genetic Programming Theory And Practice Vii,2010:119-134.
[11] Katrina Clay. Program Trading Averaged28.6Percent of NYSE Volumeduring May16-20[M/OL].2011,(6).\\http://www. nyse. com/press/1302173178545. html.
[12]杨喆.全球顶尖程序化交易模型研究[R].量化研究专题报告,海通证券,2012,(2).
[13]闫冀楠,张维,刘志弘.金融技术发展的智能化趋势-人工智能及其杂合对当代金融理论与实践的影响[J].河南金融管理干部学院学报,1998,(5):14-16.
[14] Anis A. A., Llogy E. H. The Expected Value of the Adjusted RescaledHurst Range of Independent National Summands. Biometrical Journal.Black F, Scholes M. The Pricing of Options and Corporate Liabilities. TheJournal of Political Economy,1996,(3).
[15] Daniela Federici, Giancarlo Gandlfo. Chaos and The Exchange Rate.Journal oF International Trade and Economic Development,2002,(5).
[16] Hai Chin Yu, Ming Chang Huang. Statistical Properties of Volatility inFractal Dimension and Probability Distribution among Six Stock Markets.Applied Financial Economics,2004,(7).
[17] Klaasjan Visscher,, Arie Rip. Coping with Chaos in Change Processes.Creativity and Innovation Management,2003.
[18] Richard H,Day, Oleg V. Pavlov. Computing Economic Chaos[J].Computational Economics,2004,(4).
[19]彭晓东,莫东松,刘勇,夏新初,夏克俭.基于数据仓库的综合决策支持系统的设计研究[J].计算机工程与设计,2003,(5):33-36.
[20]冉宇瑶.证券数据仓库在证券业的应用及其结构[J].计算机工程与设计,2004,(8):45-48.
[21]温小敏,顾锋娟.马尔可夫交易策略研究[J].统计与决策,2007,(1).
[22]梁元星.测股市分析股价的随机过程模型的建构[J].广西民族学院学报,(自然科学版),2003,(3).
[23] Brock, Lakonishok, LeBaron. Simple Technical Trading Rules and TheStochastic Properties of Stock Returns[J]. The Journal of Finance XLVII,1992,(5):1731-1764.
[24]周明磊.非参数估计与小波分析在股市趋势线中的应用[J].数理统计与管理,2005,(4).
[25]高雷,任慧玉.基于小波分析的我国股票市场趋势探测[J].统计与决策,2005,(16).
[26] Chan, Jegadeesh, Lakonishok. Momentum Strategies[J]. The Journal ofFinance,1996,(5):1681-1713.
[27] Hill T, Connor M, Remus, W. Neural Network Models for Time SeriesForecasts[J]. Management Science,1996.
[28] Jun Yu. Forecasting Volatility in The New Zealand Stock Market[J]. Journalof Fianancial Economics,2002.
[29] Tao Wang. Based on the Theory of Fuzzy of Stock Evaluation System[J],2008.
[30] NaiYang Deng, YingJie Tie. The New Method on Data Mining—theSupport Vector Machine[J],2004.
[31] NaiYang Deng, YingJie Tie. Support Vector Machine-theory[J]. Algorithmand the Development,2009.
[32] Bo Lu. The Application of Support Vector Machine in the Classification[J].International Journal of Image and Graphics,2005.
[33] Wei Sheng. Model Research of Stock Price Prediction[J]. EconomicProblems Research,2009.
[34] Neftci., Nanve Trading Rules in Financial Markets and Wiener KolmogorovPrediction Theory: A Study of Technical Analysis[J]. Journal of Business,1991,(64):549-571.
[35] Nez-Letamendia. Fitting the Control Parameters of a Genetic Algorithm: anApplication to Technical Trading Systems Design[J]. European Journal ofOperational Research,2007,(179):847-868.
[36] Andrew. Szakmary, Qian Shen, Subhash. Trend-following TradingStrategies In Commodity Futures: A re-examination[J]. Journal of Banking&Finance,2010,(34):409-426.
[37]王庆宗.移动平均线建构动量策略研究[J].河南科技,2010,(14):21-23.
[38]苏匡,罗元,方楚贤.基于神经网络对我国股票市场技术分析的实证检验[J].当代经济,2010,(7):146-147.
[39]尹宇明.中国股市技术交易规则有效性的计量分析[J].现代财经(天津财经大学学报),2011,(6):81-87.
[40]韦增欣,周智超,韦鑫.技术分析在A股市场的有效性研究及实证检验[J].重庆理工大学学报(自然科学),2013,(7):113-120.
[41]许红伟,吴冲锋,张翔.我国量化（对冲）基金业绩表现的比较分析[J].投资研究,2013,(2):72-87.
[42] Dourra, Siy. Investment using Technical Analysis and Fuzzy Logic[J].Fuzzy sets and Systems,2002,(127):221-240.
[43] George S. Atsalakis, Kimon P. Valavanis. Surveying Stock MarketForecasting Techniques–Part II: Soft Computing Methods[J]. ExpertSystems with Applications,2009,(36):5932-5941.
[44] L. Dymova, P. Sevastianov, P. Bartosiewicz.A New Approach to TheRule-base Evidential Reasoning: Stock Trading Expert System Application[J].Expert Systems with Applications,2010,(37):5564-5576.
[45] Pei-Chann Chang, Chin-Yuan Fan, Jun-Lin Lin.Trend Discovery inFinancial Time Series Data using a Case Based fuzzy Decision Tree[J].Expert Systems with Applications,2011,(38):6070-6080.
[46]张杨,宋恒.基于聚类技术的股市基本趋势规律挖掘[J].数理统计与管理,2006,(7):18-21.
[47]汪东.基于支持向量机的选股与选时研究[J].博士毕业论文,上海交通大学,2007.
[48]赵尔波,马欢,韩战钢.遗传程序设计分析股价移动平均及中长期走势[J].计算机应用研究,2010,(6):17-18.
[49]谢丁.反趋向类技术分析指标的实证研究及评测[J].黑龙江金融,2011,(3):38-40.
[50]苏治,傅晓媛.核主成分遗传算法与SVR选股模型改进[J].统计研究,2013,(5):54-62.
[51] Pruitt, White R. The CRISMA Trading System: Who Says TechnicalAnalysis Can’t Beat the Market?[J]. Journal of Portfolio Management,1988,(1):55-58.
[52] Potvin, Soriano, Vallte. Generating Trading Rules on the Stock Marketswith Genetic Programming[J]. Computers&Operations Research,2004,(31):1033-1047.
[53] Kwon, Moon. A Hybrid Neurogenetic Approach for Stock Forecasting[J].IEEE Transactions on Neural Networks,2007,18(3):851-864.
[54] Abeyratna, David. The Profitability of Moving Average Trading[J]. Rulesin South Asian Stock Markets. Emerging Markets Review,2001:17-33.
[55] Laura Nu′n ez-Letamendia.Fitting the Control Parameters of a GeneticAlgorithm: An Application to Technical Trading Systems Design[J].European Journal of Operational Research,2007,(179):847-868.
[56] Casqueiro, Rodrigues L. Neuro-dynamic Trading Methods[J]. EuropeanJournal of Operational Research,2006,(175):1400-1412.
[57] Lai Ming-Ming. The Profitability of the Simple Moving Averages andTrading Range Breakout in the Asian Stock Market[J]. Journal of AsianEconomics,2006,(17):144-170.
[58] Stephanos Papadamou, George Stephanides. Improving Technical TradingSystems by Using a New MATLAB-based Genetic Algorithm Procedure[J].Mathematical and computer Modelling,2007,(46):189-197.
[59] Andrada-ftlix, Fernandez-Rodríguez. Improving Moving Average TradingRules with Boosting and Statistical Learning Methods[J]. Journal ofForecasting,2008,(27):433-449.
[60] Thira Chavar nakul, David Enke. Intelligent Technical Analysis Based EquiVolume Charting for Stock Trading using Neural Networks[J]. ExpertSystems with Applications,2008,(34):1004-1017.
[61] Pei-Chann Changa, T. Warren Liaob, Jyun-Jie Lina, Chin-Yuan Fanc.ADynamic Threshold Decision System for Stock Trading Signal Detection[J].Applied Soft Computing,2011,(11):3998-4010.
[62] Massimiliano. Investment using Evolutionary Learning Methods andTechnical Rules[J]. European Journal of Operational Research207,2010:1717-1727.
[63] Hoi Tin Kong, Qing Zhang, G. George Yin. A Trend-following Strategy:Conditions for Optimality[J]. Automatica,2011,(47):661-667.
[64]王兆军,曾渊沧,郝刚.移动平均线方法的最佳步长组合的确定[J].高校应用数学学报A,2000,(2):219-228.
[65]曾宪聚,代文强,冯建栋.基于移动平均线的证券投资决策方法研究[J].经济师,2003,(10):134-136.
[66]刘厉兵.基于马尔可夫链的移动平均线分析新思维[J].商业研究,2005,(12):137-138.
[67]丁圣,高风.小波神经网络在股票平均线交易规则中的应用[J].计算机仿真,2006,(11):259-262.
[68]简清明,曾黄麟.基于移动平均线组合和支持向量机的股市趋势研究[J].计算机应用与软件,2011,(12):81-83.
[69]张伟,李泓仪,兰书梅,张洁. GA-SVM对上证综指走势的预测研究[J].东北师大学报（自然科学版）,2012,(3):55-59.
[70] Allen, Karjalainen. Using Genetic Algorithms to Find Technical TradingRules[J]. Journal of Financial Economics,1999,(51):245-271.
[71] Stephan Schulmeister.Profitability of technical Stock Trading: HAS itMoved From daily to Intraday Data?[J]. Review of Financial Economics,2009,(18):190-201.
[72] Lamartine Almeida Teixeira, Adriano Lorena Inácio de Oliveira.A methodfor Automatic Stock Trading Combining Technical Analysis and NearestNeighbor Classification[J]. Expert Systems with Applications,2010,(37):6885-6890.
[73] Simon,Yan. Trend Following Algorithms In Automated Derivatives MarketTrading[J]. Expert Systems with Applications,2012,(4):7758-7771.
[74]苏涛.金融市场风险VaR度量方法的改进研究.天津大学博士论文,2007.
[75]郑文通.金融风险管理的VAR方法及其应用[J].国际金融研究,1997,(9):76-79.
[76]杨春鹏.金融衍生工具风险管理的概率方法[J].数量经济技术经济研究,1998,(11):36-39.
[77]王春峰,蒋祥林,李刚.基于随机波动性模型的中国股市波动性估计[J].管理科学学报,2003,(4):8-12.
[78]马超群,李红权,徐山鹰,杨晓光,李晖.风险价值的完全参数方法及其在金融市场风险管理中的应用[J].系统工程理论与实践,2001,(4):44-48.
[79]江红莉,何建敏,胡小平.基于时变Copula的La-VaR测度研究[J].重庆大学学报,(社会科学版),2013,(3):27-33.
[80]庄新路,庄新田,黄小原.基于VAR风险指标的投资组合模糊优化[J].数学的实践与认识,2003,(3):55-57.
[81]陈立新. VaR风险测量模型在我国股票市场中的应用[J].大连铁道学院学报,2004,(2):44-48.
[82]蒋虹,曲丹丹.基于VaR的沪深300股指期货风险管理实证研究[J].经济问题,2008,(12):74-80.
[83]张鹏.不允许卖空情况下均值-方差和均值-VaR投资组合比较研究[J].中国管理科学,2008,(4):64-69.
[84]吴中冉,徐强.基于VaR的股指期货套期保值研究[J].金融经济,2010,(8):61-64.
[85]钟纯,吴雪,陈文财.基于VaR和CVaR下优化模型及其在保险资金组合投资中的应用[J].南昌大学学报,(工科版),2011,(4):398-403.
[86]魏波.多因素证券投资组合在VaR风险约束下的模型及实证研究[J].长春工业大学学报,(社会科学版),2013,(4):29-32.
[87]范K撤普.通向金融王国之路[M].机械工业出版社,1998:233-246.
[88]刘伟,杨廷干.关于股市技术分析的风险测度研究[J].金融与经济,2010,(9):57-59.
[89] Tushar S. Chande.超越技术分析[M].万卷出版公司,2010:20-25.
[90]丁鹏.量化投资[M].电子工业出版社,2012:20-23.
[91] Philip Treleaven, Michal Galas, Vidhi Lalc hand. Algorithmic Trading Review[J]. Communications of The Acm,2013,(9):76-85.
[92]奥尔德里奇.打开量化投资的黑箱[M].机械工业出版社,2013:57-61.
[93]张学工.关于统计学习理论与支持向量机[J].自动化学报,(1):32-42.
[94] HsuC, LinC. A Comparison of Methods for Multiclass Support VectorMachines [J]. IEEE Transon Neural Networks,2002,13(2):415-425.
[95] Takahash i F, Abe S. Decision-Tree-Based Multiclass Support VectorMachines[A]. Procof the9th Int Confon Neural Information P rocessing[C].Singapore,2002,(3):1418-1422.
[96] Sungmoon C, Sang H O, Soo-Young L. Support Vector Machines withBinary Tree Arch Itecture for Multi-Class Classification[J]. NeuralInformation Processing-letters and Reviews,2004,2(3):47-51.
[97]唐发明,王仲东,陈绵云.支持向量机多类分类算法研究[J].控制与决策,2005,(7):746-754.
[98][98]Yao X, Liu Y. Search Methodologies. Springer[J]. Chapter MachineLearning,2005:341-373.
[99] T.Amahroq, A. Taa. On Lagrange Kuhn—Tucker Multipliers for MultiObjective Optimization Problems[J]. Optimization,1997,41:159-172.
[100]C oello, Lamont, Veldhuizen. Evolutionary Algorithms for Solving Multi-Objective Problem[J]. Springer,2007.
[101]菲利普,乔瑞.风险价值VaR—金融风险管理新标准[M].北京:中信出版社,2005:251-277.
[102]R uey S. Tsay.金融时间序列分析[M].人民邮电出版社,2009:71-127.
[103]F ishman, G. Monte Carlo: ConCept, Algorithms, and Applieations[J].Springer Series in Operational Researeh. Springer, Berlin,1997.
© 2004-2018 中国地质图书馆版权所有 京ICP备05064691号 京公网安备11010802017129号 地址：北京市海淀区学院路29号 邮编：100083 电话：办公室：(+86 10)66554848；文献借阅、咨询服务、科技查新：66554700 |