A combination of artificial neural network and random walk models for financial time series forecasting

Properly comprehending and modeling the dynamics of financial data has indispensable practical importance. The prime goal of a financial time series model is to provide reliable future forecasts which are crucial for investment planning, fiscal risk hedging, governmental policy making, etc. These time series often exhibit notoriously haphazard movements which make the task of modeling and forecasting extremely difficult. As per the research evidence, the random walk (RW) is so far the best linear model for forecasting financial data. Artificial neural network (ANN) is another promising alternative with the unique capability of nonlinear self-adaptive modeling. Numerous comparisons of the performances of RW and ANN models have also been carried out in the literature with mixed conclusions. In this paper, we propose a combination methodology which attempts to benefit from the strengths of both RW and ANN models. In our proposed approach, the linear part of a financial dataset is processed through the RW model, and the remaining nonlinear residuals are processed using an ensemble of feedforward ANN (FANN) and Elman ANN (EANN) models. The forecasting ability of the proposed scheme is examined on four real-world financial time series in terms of three popular error statistics. The obtained results clearly demonstrate that our combination method achieves reasonably better forecasting accuracies than each of RW, FANN and EANN models in isolation for all four financial time series.     

Simulating and modelling the DAX index and the USO Etf financial time series by using a simple agent-based learning architecture

This work presents an extensive case study on modelling the DAX (Deutscher Aktienindex) index and United States Oil Fund (USO) exchange-traded fund (Etf) time series with the financial agent-based system learning financial agent-based simulator (L-FABS) that exploits simulated annealing as a learning method. The USO Etf time series is highly correlated with oil price behaviour, and the DAX index is based on the weighted and accumulated behaviour of the share prices of some of the largest companies traded on the Frankfurt Stock Exchange. These two time series are driven by completely different economic factors and thus provide two diverse empirical settings to evaluate the effectiveness of our methodology. Our experimentation shows that a relatively simple computational representation of real financial markets is effective in capturing the overall behaviour of the time series with varying approximation levels while the prediction target is moved into the future. The reported experimental investigation of L-FABS shows that it is robust notwithstanding the learning method used and the data sets exploited. L-FABS indeed produced a relatively low approximation error in several settings even when evaluated with respect to other modelling approaches, for example, 0.88% and 1.61% errors on average for 1 day ahead experiments in, respectively, DAX index and USO Etf.     

Improved estimation of electricity demand function by using of artificial neural network,principal component analysis and data envelopment analysis

Due to various seasonal and monthly changes in electricity consumption and difficulties in modeling it with the conventional methods, a novel algorithm is proposed in this paper. This study presents an approach that uses Artificial Neural Network (ANN), Principal Component Analysis (PCA), Data Envelopment Analysis (DEA) and ANOVA methods to estimate and predict electricity demand for seasonal and monthly changes in electricity consumption. Pre-processing and post-processing techniques in the data mining field are used in the present study. We analyze the impact of the data pre-processing and post-processing on the ANN performance and a 680 ANN-MLP is constructed for this purpose. DEA is used to compare the constructed ANN models as well as ANN learning algorithm performance. The average, minimum, maximum and standard deviation of mean absolute percentage error (MAPE) of each constructed ANN are used as the DEA inputs. The DEA helps the user to use an appropriate ANN model as an acceptable forecasting tool. In the other words, various error calculation methods are used to find a robust ANN learning algorithm. Moreover, PCA is used as an input selection method, and a preferred time series model is chosen from the linear (ARIMA) and nonlinear models. After selecting the preferred ARIMA model, the Mcleod–Li test is applied to determine the nonlinearity condition. Once the nonlinearity condition is satisfied, the preferred nonlinear model is selected and compared with the preferred ARIMA model, and the best time series model is selected. Then, a new algorithm is developed for the time series estimation; in each case an ANN or conventional time series model is selected for the estimation and prediction. To show the applicability and superiority of the proposed ANN-PCA-DEA-ANOVA algorithm, the data regarding the Iranian electricity consumption from April 1992 to February 2004 are used. The results show that the proposed algorithm provides an accurate solution for the problem of estimating electricity consumption.     

金融时间序列模糊边界预测研究

传统的金融时间序列预测方法以精确的输入数据为研究对象,在建立回归模型的基础上做单步或多步预测,预测结果是一个或多个具体的值.由于金融市场的复杂性,传统的预测方法可靠度较低.提出将金融时间序列模糊信息粒化成一个模糊粒子序列,运用支持向量机对模糊粒子的上下界进行回归,然后应用回归所得到的模型分别对上下界进行单步预测,从而将预测的结果限定在一个范围之内.这是一种全新的思路.以上证指数周收盘指数为实验数据,实验结果表明了这种方法的有效性.     

A hybrid intelligent model based on recurrent neural networks and excitable dynamics for price prediction in deregulated electricity market

This paper examines electricity price time series from dynamical system perspective and proposes a hybrid model which employs a synergistic combination of Recurrent Neural Network (RNN) and coupled excitable system for prediction of future prices in deregulated electricity markets. Driven by profit maximizing decisions taken by various agents, these markets belong to the class of financial systems. However presence of intermittent spikes and complex dynamic nonlinearities in electricity price time series render the prediction task extremely challenging. The approximation ability of Recurrent Neural Networks to map dynamic functions together with sharp jumping attribute of coupled excitable systems allows close approximation of spiky time series. The developed hybrid model was applied for point and interval forecasting in various markets worldwide over different seasons for testing its adaptability in different environments. Satisfactory prediction results were obtained in all the markets, in stable as well as spiking regions of the time series.     

Impact of momentum bias on forecasting through knowledge discovery techniques in the foreign exchange market

To an increasing extent since the late 1980s, software learning methods including neural networks (NN) and case based reasoning (CBR) have been used for prediction in financial markets and other areas. In the past, the prediction of foreign exchange rates has focused on isolated techniques, as exemplified by the use of time series models including regression models or smoothing methods to identify cycles and trends. At best, however, the use of isolated methods can only represent fragmented models of the causative agents, which underlie business cycles. Experience with artificial intelligence applications since the early 1980s points toward a multistrategy approach to discovery and prediction.This paper investigates the impact of momentum bias on forecasting financial markets through knowledge discovery techniques. Different modes of bias are used as input into learning systems using implicit knowledge representation (NNs) and CBR. The concepts are examined in the context of predicting movements in the Japanese yen.     

Optimization of neural network model using modified bat-inspired algorithm

The success of an artificial neural network (ANN) strongly depends on the variety of the connection weights and the network structure. Among many methods used in the literature to accurately select the network weights or structure in isolate; a few researchers have attempted to select both the weights and structure of ANN automatically by using metaheuristic algorithms. This paper proposes modified bat algorithm with a new solution representation for both optimizing the weights and structure of ANNs. The algorithm, which is based on the echolocation behaviour of bats, combines the advantages of population-based and local search algorithms. In this work, ability of the basic bat algorithm and some modified versions which are based on the consideration of the personal best solution in the velocity adjustment, the mean of personal best and global best solutions through velocity adjustment and the employment of three chaotic maps are investigated. These modifications are aimed to improve the exploration and exploitation capability of bat algorithm. Different versions of the proposed bat algorithm are incorporated to handle the selection of the structure as well as weights and biases of the ANN during the training process. We then use the Taguchi method to tune the parameters of the algorithm that demonstrates the best ability compared to the other versions. Six classifications and two time series benchmark datasets are used to test the performance of the proposed approach in terms of classification and prediction accuracy. Statistical tests demonstrate that the proposed method generates some of the best results in comparison with the latest methods in the literature. Finally, our best method is applied to a real-world problem, namely to predict the future values of rainfall data and the results show satisfactory of the method.     

Predicting the direction,maximum, minimum and closing prices of daily Bitcoin exchange rate using machine learning techniques

Bitcoin is the most accepted cryptocurrency in the world, which makes it attractive for investors and traders. However, the challenge in predicting the Bitcoin exchange rate is its high volatility. Therefore, the prediction of its behavior is of great importance for financial markets. In this way, recent studies have been carried out on what internal and/or external Bitcoin information is relevant to its prediction. The increased use of machine learning techniques to predict time series and the acceptance of cryptocurrencies as financial instruments motivated the present study to seek more accurate predictions for the Bitcoin exchange rate. In this way, in a first stage of the proposed methodology, different feature selection techniques were evaluated in order to obtain the most relevant attributes for the predictions. In the sequence, it was analyzed the behavior of Artificial Neural Networks (ANN), Support Vector Machines (SVM) and Ensemble algorithms (based on Recurrent Neural Networks and the k-Means clustering method) for price direction predictions. Likewise, the ANN and SVM were employed for regression of the maximum, minimum and closing prices of the Bitcoin. Moreover, the regression results were also used as inputs to try to improve the price direction predictions. The results showed that the selected attributes and the best machine learning model achieved an improvement of more than 10%, in accuracy, for the price direction predictions, with respect to the state-of-the-art papers, using the same period of information. In relation to the maximum, minimum and closing Bitcoin prices regressions, it was possible to obtain Mean Absolute Percentage Errors between 1% and 2%. Based on these results, it was possible to demonstrate the efficacy of the proposed methodology when compared to other studies.     

Combining nonlinear independent component analysis and neural network for the prediction of Asian stock market indexes

With the economic successes of several Asian economies and their increasingly important roles in the global financial market, the prediction of Asian stock markets has becoming a hot research area. As Asian stock markets are highly dynamic and exhibit wide variation, it may more realistic and practical that assumed the stock indexes of Asian stock markets are nonlinear mixture data. In this research, a time series prediction model by combining nonlinear independent component analysis (NLICA) and neural network is proposed to forecast Asian stock markets. NLICA is a novel feature extraction technique to find independent sources from observed nonlinear mixture data where no relevant data mixing mechanisms are available. In the proposed method, we first use NLICA to transform the input space composed of original time series data into the feature space consisting of independent components representing underlying information of the original data. Then, the ICs are served as the input variables of the neural network to build prediction model. Among the Asian stock markets, Japanese and China’s stock markets are the biggest two in Asia and they respectively represent the two types of stock markets. Therefore, in order to evaluate the performance of the proposed approach, the Nikkei 225 closing index and Shanghai B-share closing index are used as illustrative examples. Experimental results show that the proposed forecasting model not only improves the prediction accuracy of the neural network approach but also outperforms the three comparison methods. The proposed stock index prediction model can be therefore a good alternative for Asian stock market indexes.     

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.     

Exploring the major factors affecting fly-ash concrete carbonation using artificial neural network

Concrete carbonation is one of the main causes of corrosion of the reinforcement and consequently causing damage to the reinforced concrete structures. The progress of the carbonation front depends on many factors including mixture proportions and exposure conditions. Several carbonation prediction models including mathematical and analytical predictions are available. Most of these models, however, are based on simple regression equations and cannot predict or accurately reflect the various factors involved in concrete carbonation. The current published results in this issue are in conflict. In view of this, our research aims to apply an artificial neural network (ANN) approach for predicting the carbonation of fly-ash concrete taking into account the most influential parameters, including mixture proportions and exposure conditions. Six parameters were considered as inputs to the ANN model, covering, binder and fly-ash content, water-to-binder ratio, CO₂ concentration, relative humidity, and time of exposure; one output is carbonation depth. The ANN model was prepared, trained, and tested with 300 datasets from experiments as well as past research. The performance of training, validation, and test sets shows a high correlation between the experimental and the ANN predicted values of the carbonation depth. In addition, the proposed prediction model was in good agreement with the experimental data in comparison with other model. This study concludes that the use of this model for numerical investigations on the parameters affecting the carbonation depth in fly-ash concrete is successful and provides scientific guidance for durability design.

     

A moving-average filter based hybrid ARIMA–ANN model for forecasting time series data

A suitable combination of linear and nonlinear models provides a more accurate prediction model than an individual linear or nonlinear model for forecasting time series data originating from various applications. The linear autoregressive integrated moving average (ARIMA) and nonlinear artificial neural network (ANN) models are explored in this paper to devise a new hybrid ARIMA–ANN model for the prediction of time series data. Many of the hybrid ARIMA–ANN models which exist in the literature apply an ARIMA model to given time series data, consider the error between the original and the ARIMA-predicted data as a nonlinear component, and model it using an ANN in different ways. Though these models give predictions with higher accuracy than the individual models, there is scope for further improvement in the accuracy if the nature of the given time series is taken into account before applying the models. In the work described in this paper, the nature of volatility was explored using a moving-average filter, and then an ARIMA and an ANN model were suitably applied. Using a simulated data set and experimental data sets such as sunspot data, electricity price data, and stock market data, the proposed hybrid ARIMA–ANN model was applied along with individual ARIMA and ANN models and some existing hybrid ARIMA–ANN models. The results obtained from all of these data sets show that for both one-step-ahead and multistep-ahead forecasts, the proposed hybrid model has higher prediction accuracy.     

一种面向金融数据的基于ARIMA的模糊时间序列预测模型

时序分析方法在金融数据挖掘中扮演着越来越重要的角色,然而,历史数据的不完整、不确切性制约着传统金融时间序列预测方法的准确性。创新地定义ARIMA模型的相似性和模,并融合模糊时间序列方法,提出新的基于ARIMA的模糊时间序列预测模型。该模型能够高效处理不完整的、含糊的历史数据,并对未来走势进行有效预测。一方面, ARIMA模型的简约灵活性使得对高维金融时间序列的特征提取大为简化;另一方面,由于结合模糊逻辑的理论,该方法能够有效发现历史数据中的相似模式。以人民币兑美元汇率为例,通过对预测结果的分析,验证了的新模型的有效性。     

Forecasting time series using a methodology based on autoregressive integrated moving average and genetic programming

The autoregressive integrated moving average (ARIMA), which is a conventional statistical method, is employed in many fields to construct models for forecasting time series. Although ARIMA can be adopted to obtain a highly accurate linear forecasting model, it cannot accurately forecast nonlinear time series. Artificial neural network (ANN) can be utilized to construct more accurate forecasting model than ARIMA for nonlinear time series, but explaining the meaning of the hidden layers of ANN is difficult and, moreover, it does not yield a mathematical equation. This study proposes a hybrid forecasting model for nonlinear time series by combining ARIMA with genetic programming (GP) to improve upon both the ANN and the ARIMA forecasting models. Finally, some real data sets are adopted to demonstrate the effectiveness of the proposed forecasting model.     

Improving artificial neural networks’ performance in seasonal time series forecasting

In this study, an artificial neural network (ANN) structure is proposed for seasonal time series forecasting. The proposed structure considers the seasonal period in time series in order to determine the number of input and output neurons. The model was tested for four real-world time series. The results found by the proposed ANN were compared with the results of traditional statistical models and other ANN architectures. This comparison shows that the proposed model comes with lower prediction error than other methods. It is shown that the proposed model is especially convenient when the seasonality in time series is strong; however, if the seasonality is weak, different network structures may be more suitable.     

Modelling sovereign credit ratings: Neural networks versus ordered probit

Sovereign credit ratings are becoming increasingly important both within a financial regulatory context and as a necessary prerequisite for the development of emerging capital markets. Using a comprehensive dataset of rating agencies and countries over the period 1989–1999, this paper demonstrates that artificial neural networks (ANN) represent a superior technology for calibrating and predicting sovereign ratings relative to ordered probit modelling, which has been considered by the previous literature to be the most successful econometric approach. ANN have been applied to classification problems with great success over a wide range of applications where there is an absence of a precise theoretical model to underpin the relationships in the data. The results for sovereign credit ratings presented here corroborate other researchers' findings that ANN are highly effective classifiers.     

Predicting direction of stock price index movement using artificial neural networks and support vector machines: The sample of the Istanbul Stock Exchange

Prediction of stock price index movement is regarded as a challenging task of financial time series prediction. An accurate prediction of stock price movement may yield profits for investors. Due to the complexity of stock market data, development of efficient models for predicting is very difficult. This study attempted to develop two efficient models and compared their performances in predicting the direction of movement in the daily Istanbul Stock Exchange (ISE) National 100 Index. The models are based on two classification techniques, artificial neural networks (ANN) and support vector machines (SVM). Ten technical indicators were selected as inputs of the proposed models. Two comprehensive parameter setting experiments for both models were performed to improve their prediction performances. Experimental results showed that average performance of ANN model (75.74%) was found significantly better than that of SVM model (71.52%).     

A new hybrid for improvement of auto-regressive integrated moving average models applying particle swarm optimization

A time series forecasting is an active research applied significantly in a variety of economics areas. Over the past three decades an auto-regressive integrated moving average (ARIMA) model, as one of the most important time series models, has been applied in financial markets forecasting. Recent researches in time series forecasting ARIMA models indicate some basic limitations which detract from their popularities for financial time series forecasting: One limitation of an ARIMA model is that it requires a large amount of historical data to generate an accurate result. Both theoretical and empirical findings suggest that combining different time series models may be an effective method of improving the predictive performances of data especially when the models in the ensemble are quite different. The main purpose of present paper is to combine the ARIMA model with the particle swarm optimization (PSO) model in order to improve and generate more accurate forecasting results. Under small data information, combining the PSO and ARIMA models performs better performance results compared to an ARIMA model itself. The proposed model is robust and it may be used as an alternative forecasting tool in economics areas.     

Volatility forecast using hybrid Neural Network models

In this research the testing of a hybrid Neural Networks-GARCH model for volatility forecast is performed in three Latin-American stock exchange indexes from Brazil, Chile and Mexico. A detail of the methodology and application of the volatility forecast of financial series using a hybrid artificial Neural Network model are presented.The results demonstrate that the ANN models can improve the forecasting performance of the GARCH models when studied in the three Latin-American markets and it is shown that the results are robust and consistent for different ANN specifications and different volatility measures.     

A combined neural-wavelet model for prediction of Ligvanchai watershed precipitation

Doubtlessly the first step in a river management is the precipitation modeling over the related watershed. However, considering high-stochastic property of the process, many models are being still developed in order to define such a complex phenomenon in the field of hydrologic engineering. Recently artificial neural network (ANN) as a non-linear inter-extrapolator is extensively used by hydrologists for rainfall modeling as well as other fields of hydrology.In the current research, the wavelet analysis was linked to the ANN concept for prediction of Ligvanchai watershed precipitation at Tabriz, Iran. For this purpose, the main time series was decomposed to some multi-frequently time series by wavelet theory, then these time series were imposed as input data to the ANN to predict the precipitation 1 month ahead. The obtained results show the proposed model can predict both short- and long-term precipitation events because of using multi-scale time series as the ANN input layer.