基于深度强化学习的金融交易算法研究

交易策略在金融资产交易中具有十分重要的作用,如何在复杂动态金融市场中自动化选择交易策略是现代金融重要研究方向。强化学习算法通过与实际环境交互作用,寻找最优动态交易策略,最大化获取收益。提出了一个融合了CNN与LSTM的端到端深度强化学习自动化交易算法,CNN模块感知股票动态市场条件以及抽取动态特征,LSTM模块循环学习动态时间序列规律,最后通过强化学习方法累积最终收益并做出交易策略。在真实股票数据上的实验结果表明,该方法显著优于基准方法,可扩展性更强,鲁棒性更好。     

一种基于收益率相对波动的股票配对交易系统的设计与实现

配对交易是统计套利的一个部分。当前配对交易对于股票对的选取大都基于统计学相关性理论,股票对的选取理论与交易对冲系统的策略无关联。主要介绍一种基于累计收益的配对方法,该方法使用累计收益的价差(相对波动)为依据来考察股票之间的相关性,通过历史收益率来确定股票之间的相关性。该方法的目标是:找到相关程度高且股价相对波动幅度大的股票对,在股价波动过程中反复套利,从而达到赢利的目的。经过模拟交易测试发现,交易策略在上证指数大涨时收益很好,在上证指数震荡和下跌时,收益依然令人满意。     

基于模糊近似支持向量回归的股价预测研究

股价预测是投资策略形成和风险管理模型发展的基础。为了降低股价变化趋势中的噪声信息和投资者关于两种股价预测误差的不同偏好对股价预测的影响,提出了基于信噪比的模糊近似支持向量回归(FPSVR)的股价预测模型。首先构建信噪比输入变量,然后引入模糊隶属度和双边权重测量方法对支持向量回归(SVR)模型进行改进,最后借助沪深300成份股2008至2019年的股票时间序列日数据,按照股市的波动情况将其分为三个阶段(牛市、熊市、震荡市),并建立三个基准模型进行对比分析。研究结果表明:与三个基准模型相比,所提出的股价预测模型的预测误差最低;与原有的SVR模型相比,FPSVR模型可以更好地对处于牛市和震荡市阶段的股票时间序列进行股价预测。     

基于MCPDDPG的智能车辆路径规划方法及应用

针对智能车路径规划过程中常存在动态环境感知预估不足的问题,使用基于蒙特卡罗深度策略梯度学习(Monte Carlo prediction deep deterministic policy gradient, MCPDDPG)的智能车辆路径规划方法,设计一种基于环境感知预测、行为决策和控制序列生成的框架,实现实时的决策和规划,并输出连续的车辆控制序列.首先,利用序贯蒙特卡罗预估他车行为状态量;然后,设计基于强化Q学习的行为决策方法,使智能车辆实时预知碰撞风险,采取合理的规避策略;最后,构建深度策略梯度学习网络框架,获取智能车辆规划路径的最优轨迹序列.实验结果表明,所提方法能够缓解环境感知的预估不足问题,提升智能车辆行为决策的快速性,保障路径规划的主动安全,并输出连续的轨迹序列,为智能车辆导航控制提供前提.     

基于时序超图卷积神经网络的股票趋势预测方法

传统的股票预测方法大多基于时间序列模型,忽视了股票之间复杂的关系,并且该关系往往超出成对连接,例如同行业板块内股票或者基金持仓多支股票。针对该问题,提出一种基于时序超图卷积神经网络（HGCN）的股价走势预测方法,根据金融投资事实构造超图模型以拟合股票之间的多元关系,该模型包括两大组件：门控循环单元（GRU）网络和超图卷积神经网络。GRU网络对历史数据进行时间序列建模,捕捉长期依赖关系;HGCN建模股票间的高阶关系以学习内在关系属性,从而将股票间多元关系信息引入到传统的时序建模中,进行端到端的趋势预测。在中国A股市场真实数据集上的实验结果表明,相较于已有的股票预测方法,所提模型预测性能有所提升;如与GRU网络相比,所提模型在ACC和F1_score上的相对增幅分别为9.74%和8.13%,且更具有稳定性。此外,模拟回测结果显示,基于该模型的交易策略更具获利能力,年回报率达到11.30%,与长短期记忆（LSTM）网络相比提高了5个百分点。     

一种多源数据驱动的自动交易系统决策模型

股票自动交易系统属于典型的复杂系统,其成功的关键是如何对股价进行有效的预测与决策.股价受多种信息的影响,但传统的自动交易模型多建立在历史交易数据的基础上.针对上述问题,本文综合利用新闻文本数据与股价技术指标数据,基于人工神经网络（Artificial neural netuorks,ANN）方法设计了一种多源数据驱动的股票自动交易决策模型.本文首先分析了各类财经新闻的特点及其对股价的影响,然后设计了相应模板抽取了中文文本中的财经新闻事件;在此基础上,设计了历史股价和新闻事件数据共同驱动的ANN-News模型,并利用实际数据验证了模型的有效性.实验发现,ANN-News模型比传统的机器学习类模型股价预测准确率提升约4%,收益率提升约7%.     

面向排名预测的上市公司股价收益研究

对于在深圳证券交易所上市的公司,通过分析和挖掘其季度报表或者相关交易网站中的数据,提取到排名预测任务中相关的数据特征以及通过爬虫获得的文本特征,成功构建了公司每股收益预测排名的模型,实现了对股价收益排名的合理预测.实验结果表明,我们的提出的模型能够有效的提高股价排名预测任务的性能,其中SPRP-Random Forests模型在NDCG@10评价指标中可以达到0.9583.在为股民选择股票,公司经营模式调整等方面具有一定的实用价值.     

基于深度跨模态信息融合网络的股票走势预测

股票走势预测是经典且具有挑战性的任务，可帮助交易者做出获得更大收益的交易决策。近年来，基于深度学习的股票走势预测方法的性能得到明显提升，但现有方法大多仅依托于股票价格的历史数据来完成走势预测，无法捕捉价格指标之外的市场动态规律，在一定程度上限制了方法的性能。为此，将社交媒体文本与股票历史价格信息相结合，提出了一种基于深度跨模态信息融合网络(DCIFNet)的股票走势预测新方法。DCIFNet首先采用时间卷积操作对股票价格和推特文本进行编码，使得每个元素对其邻域元素都有足够的了解；然后，将结果输入到基于transformer的跨模态融合结构中，以更有效地融合股票价格和推特文本中的重要信息；最后，引入多图卷积注意力网络从不同角度描述不同股票之间的相互关系，能够更有效地捕获关联股票间的行业、维基和相关关系，从而提升股票走势预测的精度。在9个不同行业的高频交易数据集上实施走势预测和模拟交易实验。消融实验及所提方法与用于股票预测的多管齐下的注意力网络(MAN-SF)方法的比较结果验证了DCIFNet方法的有效性，准确率达到了0.630 9,明显优于领域内代表性方法。     

基于MODWT在金融数据预测的应用

为了准确的把握股价的趋势走向,提出了一种基于极大重叠离散小波变换(MODWT)时间序列分析的股价预测方法(M-ARMA).该方法是对股价时间序列利用mallat算法对其进行极大重叠离散小波变换,使得整个序列分解成不同频率的序列,同时利用小波分析在时域和频域上都具有良好的局部化性质,多尺度分析功能,结合ARMA模型的预测方法,以较为准确地根据历史数据预测其将来短期的走势.实验表明,MODWT时间序列分析方法比传统的时间序列分析方法预测的精度更高.     

基于Pytorch的LSTM模型对股价的分析与预测

股价预测一直都是股票投资者重点关注和重点研究的方向,针对股价具有高度非线性、高噪声、动态性等问题,提出一种基于自组织特征映射(SOM)神经网络和长短期记忆网络(LSTM)共同应用的股价预测方法。第一步聚类,使用python语言实现改进的自组织特征映射神经网络算法,将187支股票分成三类,三类股票以盈利能力大小进行聚类,并且求出每一类所包含的股票代码;第二步预测,基于Pytorch深度学习框架构造长短期记忆网络模型,分别对每一类中随机的3支股票进行股价预测,再通过均方误差和决定系数对预测结果进行评价。结果表明,在使用相同的预测模型对不同盈利能力的股票做股价预测时,盈利能力越大的股票,预测精度越高。此研究可以为投资者筛选出盈利能力更大的股票,并且在提高股价预测精度上也具有一定的贡献。     

基于神经网络集成学习算法的金融时间序列预测

本文在传统神经网络(NN)、循环神经网络(RNN)、长短时记忆网络(LSTM)与门控循环单元(GRU)等神经网络时间预测模型基础上, 进一步构建集成学习(EL)时间序列预测模型, 研究神经网络类模型、集成学习模型和传统时间序列模型在股票指数预测上的表现. 本文以16只A股和国际股票市场指数为样本, 比较模型在不同预测期间和不同国家和地区股票市场上的表现.本文主要结论如下: 第一, 神经网络类时间序列预测模型和神经网络集成学习时间序列预测模型在表现上显著稳健优于传统金融时间序列预测模型, 预测性能提高大约35%; 第二, 神经网络类模型和神经网络集成学习模型在中国和美国股票市场上的表现优于其他发达国家和地区的股票市场.     

Stock Trading Using RSPOP: A Novel Rough Set-Based Neuro-Fuzzy Approach

This paper investigates the method of forecasting stock price difference on artificially generated price series data using neuro-fuzzy systems and neural networks. As trading profits is more important to an investor than statistical performance, this paper proposes a novel rough set-based neuro-fuzzy stock trading decision model called stock trading using rough set-based pseudo outer-product (RSPOP) which synergizes the price difference forecast method with a forecast bottleneck free trading decision model. The proposed stock trading with forecast model uses the pseudo outer-product based fuzzy neural network using the compositional rule of inference [POPFNN-CRI(S)] with fuzzy rules identified using the RSPOP algorithm as the underlying predictor model and simple moving average trading rules in the stock trading decision model. Experimental results using the proposed stock trading with RSPOP forecast model on real world stock market data are presented. Trading profits in terms of portfolio end values obtained are benchmarked against stock trading with dynamic evolving neural-fuzzy inference system (DENFIS) forecast model, the stock trading without forecast model and the stock trading with ideal forecast model. Experimental results showed that the proposed model identified rules with greater interpretability and yielded significantly higher profits than the stock trading with DENFIS forecast model and the stock trading without forecast model.     

Improving stock index forecasts by using a new weighted fuzzy-trend time series method

We propose using new weighted operators in fuzzy time series to forecast the future performance of stock market indices. Based on the chronological sequence of weights associated with the original fuzzy logical relationships, we define both chronological-order and trend-order weights, and incorporate our proposals for the ex-post forecast into the classical modeling approach of fuzzy time series. These modifications for the assignation of weights affect the forecasting process, because we use jumps as technical indicators to predict stock trends, and additionally, they provide a trapezoidal fuzzy number as a forecast of the future performance of the stock index value. Working with trapezoidal fuzzy numbers allows us to analyze both the expected value and the ambiguity of the future behavior of the stock index, using a possibilistic interval-valued mean approach. Therefore, using fuzzy logic more useful information is provided to the decision analyst, which should be appropriate in a financial context. We analyze the effectiveness of our approach with respect to other weighted fuzzy time series methods using trading data sets from the Taiwan Stock Index (TAIEX), the Japanese NIKKEI Index, the German Stock Index (DAX) and the Spanish Stock Index (IBEX35). The comparative results indicate the better accuracy of our procedure for point-wise one-step ahead forecasts.     

Adaptive stock trading with dynamic asset allocation using reinforcement learning

Stock trading is an important decision-making problem that involves both stock selection and asset management. Though many promising results have been reported for predicting prices, selecting stocks, and managing assets using machine-learning techniques, considering all of them is challenging because of their complexity. In this paper, we present a new stock trading method that incorporates dynamic asset allocation in a reinforcement-learning framework. The proposed asset allocation strategy, called meta policy (MP), is designed to utilize the temporal information from both stock recommendations and the ratio of the stock fund over the asset. Local traders are constructed with pattern-based multiple predictors, and used to decide the purchase money per recommendation. Formulating the MP in the reinforcement learning framework is achieved by a compact design of the environment and the learning agent. Experimental results using the Korean stock market show that the proposed MP method outperforms other fixed asset-allocation strategies, and reduces the risks inherent in local traders.     

A comparative study of artificial neural networks,and decision trees for digital game content stocks price prediction

Precise prediction of stock prices is difficult chiefly because of the many intervening factors. Unpredictability is particularly notable in the aftermath of the global financial crisis. Data mining may however be used to discover highly correlated estimation models. This study looks at artificial neural networks (ANN), decision trees and the hybrid model of ANN and decision trees (hybrid model), the three common algorithm methods used for numerical analysis, to forecast stock prices. The author compared the stock price forecasting models derived from the three methods, and applied the models on 10 different stocks in 320 data sets in an empirical forecast. Average accuracy of ANN is 15.31%, the highest, in terms of match with real market stock prices, followed by decision trees, at 14.06%; hybrid model is 13.75%. The study also discovers that compared to the other two methods, ANN is a more stable method for predicting stock prices in the volatile post-crisis stock market.     

Stock-Price Forecasting Based on XGBoost and LSTM

Using time-series data analysis for stock-price forecasting (SPF) is complex and challenging because many factors can influence stock prices (e.g., inflation, seasonality, economic policy, societal behaviors). Such factors can be analyzed over time for SPF. Machine learning and deep learning have been shown to obtain better forecasts of stock prices than traditional approaches. This study, therefore, proposed a method to enhance the performance of an SPF system based on advanced machine learning and deep learning approaches. First, we applied extreme gradient boosting as a feature-selection technique to extract important features from high-dimensional time-series data and remove redundant features. Then, we fed selected features into a deep long short-term memory (LSTM) network to forecast stock prices. The deep LSTM network was used to reflect the temporal nature of the input time series and fully exploit future contextual information. The complex structure enables this network to capture more stochasticity within the stock price. The method does not change when applied to stock data or Forex data. Experimental results based on a Forex dataset covering 2008–2018 showed that our approach outperformed the baseline autoregressive integrated moving average approach with regard to mean absolute error, mean squared error, and root-mean-square error.     

An integrated framework of genetic network programming and multi-layer perceptron neural network for prediction of daily stock return: An application in Tehran stock exchange market

Evolutionary algorithms are generally used to find or generate the best individuals in a population. Whenever these algorithms are applied to agent systems, they will lead to optimal solutions. Genetic Network Programming (GNP), which contains graph networks, is one of the developed evolutionary algorithms. When the aim is to forecast the share price or return, ascending and descending trends, volatilities, recent returns, fundamental and technical factors have remarkable impacts on the prediction. This is why technical indicators are used to constitute a set of trading rules. In this paper, we apply an integrated framework consisting of GNP model along with a reinforcement learning and Multi-Layer Perceptron (MLP) neural network to classify data and also time series models to forecast the stock return. Moreover, we utilize rules of accumulation based on the GNP model’s results to forecast the return. The aim of using these models alongside one another is to estimate one-day return. The results derived from 9 stocks with regard to the Tehran Stock Exchange Market. GNP extracts a prodigious number of rules on the basis of 5 technical indicators with 3 times period. Next, MLP network classifies data and finds the similarity between future data and past data concerning a stock (5 sub-period) through classification. Subsequently, a number of conditions are established, in order to choose the best estimation between GNP-RL and ARMA. Distinct comparison with the ARMA–GARCH model, which is operated for return estimation and risk measurement in many researches, demonstrates an extended forecasting power of the proposed model, by the name of GNP–ARMA, reducing error by a mean of 16%.     

A Multiagent Approach to Q-Learning for Daily Stock Trading

The portfolio management for trading in the stock market poses a challenging stochastic control problem of significant commercial interests to finance industry. To date, many researchers have proposed various methods to build an intelligent portfolio management system that can recommend financial decisions for daily stock trading. Many promising results have been reported from the supervised learning community on the possibility of building a profitable trading system. More recently, several studies have shown that even the problem of integrating stock price prediction results with trading strategies can be successfully addressed by applying reinforcement learning algorithms. Motivated by this, we present a new stock trading framework that attempts to further enhance the performance of reinforcement learning-based systems. The proposed approach incorporates multiple Q-learning agents, allowing them to effectively divide and conquer the stock trading problem by defining necessary roles for cooperatively carrying out stock pricing and selection decisions. Furthermore, in an attempt to address the complexity issue when considering a large amount of data to obtain long-term dependence among the stock prices, we present a representation scheme that can succinctly summarize the history of price changes. Experimental results on a Korean stock market show that the proposed trading framework outperforms those trained by other alternative approaches both in terms of profit and risk management.     

Predicting stock index increments by neural networks: The role of trading volume under different horizons

Recent studies show that there is a significant bidirectional nonlinear causality between stock return and trading volume. In this research, we reinforce this statement and the results presented in some earlier literatures and further investigate whether trading volume can significantly improve the prediction performance of neural networks under short-, medium-and long-term forecasting horizons. An application of component-based neural networks is used in forecasting one-step ahead stock index increments. The models are also augmented by the addition of different combinations of indices’ and component stocks’ trading volumes as inputs to form more general ex-ante forecasting models. Neural networks are trained with the data of stock returns and volumes from NASDAQ, DJIA and STI indices. Results indicate that augmented neural network models with trading volumes lead to improvements, at different extents, in forecasting performance under different terms of forecasting horizon. Empirical results indicate that trading volumes lead to modest improvements on the performance of stock index increments prediction under medium-and long-term horizons.