usingANNmodelsinstockmarketindexprediction.pdf资源-CSDN文库

需积分: 9 34 浏览量 2020-09-14 10:59:09 上传评论收藏 365KB PDF 举报

Guresen, E. , Kayakutlu, G. , & Daim, T. U. (2011). Using artificial neural network models in stock market index prediction. Expert Systems with Applications, 38 , 10389–10397 . 标题中的"Using Artificial Neural Network Models in Stock Market Index Prediction"指的是利用人工神经网络（ANN）模型来预测股票市场指数。在金融领域，特别是股票市场，预测指数走势是投资者、分析师和金融机构的重要任务，因为它有助于决策者制定投资策略。描述中提到的研究由Guresen、Kayakutlu和Daim进行，他们探讨了神经网络模型在股票市场指数预测中的效用，对比了多层感知机（MLP）、动态人工神经网络（DAN2）以及结合广义自回归条件异方差性（GARCH）的混合神经网络模型。文章关键词包括“Financial time series (FTS) prediction”（金融时间序列预测），“Recurrent neural networks (RNN)”（循环神经网络），“Dynamic artificial neural networks (DAN2)”（动态人工神经网络）以及“Hybrid forecasting models”（混合预测模型）。这些概念表明研究涵盖了多种类型的神经网络模型，并尝试将它们应用于复杂的时间序列数据，如股票市场的日交易数据。文章摘要指出，近年来，许多神经网络模型和混合模型被提出用于提高股票市场预测的准确性，试图超越传统的线性和非线性方法。具体来说，研究比较了MLP、DAN2和采用GARCH技术的混合神经网络模型。通过均方误差（MSE）和平均绝对偏差（MAD）这两个指标，使用NASDAQ股票交易所的真实日交易数据对这些模型进行了评估。 1. 人工神经网络（ANN）模型在股票预测中的应用： - 多层感知机（MLP）：这是一种前馈神经网络，具有至少一个隐藏层，能处理非线性关系，适用于股票市场的复杂数据。 - 动态人工神经网络（DAN2）：这类网络能够考虑时间序列数据的动态变化，更好地捕捉市场趋势。 - 混合模型：结合了GARCH模型，该模型用于描述和预测资产回报率的波动性，从而提供新的输入变量，增强预测能力。 2. 金融时间序列预测的重要性： - 预测金融市场波动有助于投资者理解市场动态，及时调整投资组合，降低风险。 - 在日益波动的金融市场和全球化资本流动背景下，准确的预测对于商业银行和投资银行的资产管理至关重要。 3. 统计评价指标： - 均方误差（MSE）：衡量模型预测值与实际值之间差异的平方和的平均值，反映了模型预测的精确度。 - 平均绝对偏差（MAD）：计算的是预测值与实际值之间的绝对差值的平均值，它不敏感于极端值，提供了另一种度量预测准确性的方法。总结，这项研究揭示了神经网络模型在股票市场指数预测中的潜力，通过比较不同模型的性能，为金融从业者提供了更多选择。然而，实际应用中，预测结果仍受多种因素影响，如市场情绪、政策变化、全球经济环境等，因此模型的预测能力应与其他分析工具结合使用，以提高决策质量。

资源推荐

资源详情

资源评论

Using artiﬁcial neural network models in stock market index prediction

Erkam Guresen

, Gulgun Kayakutlu

⇑

, Tugrul U. Daim

Istanbul Technical University, Istanbul, Turkey

Portland State University, Portland OR, USA

article info

Keywords:

Financial time series (FTS) prediction

Recurrent neural networks (RNN)

Dynamic artiﬁcial neural networks (DAN2)

Hybrid forecasting models

abstract

Forecasting stock exchange rates is an important ﬁnancial problem that is receiving increasing attention.

During the last few years, a number of neural network models and hybrid models have been proposed for

obtaining accurate prediction results, in an attempt to outperform the traditional linear and nonlinear

approaches. This paper evaluates the effectiveness of neural network models which are known to be

dynamic and effective in stock-market predictions. The models analysed are multi-layer perceptron

(MLP), dynamic artiﬁcial neural network (DAN2) and the hybrid neural networks which use generalized

autoregressive conditional heteroscedasticity (GARCH) to extract new input variables. The comparison

for each model is done in two view points: Mean Square Error (MSE) and Mean Absolute Deviate

(MAD) using real exchange daily rate values of NASDAQ Stock Exchange index.

1. Introduction

Forecasting simply means understanding which variables lead

to predict other variables (Mcnelis, 2005). This means a clear

understanding of the timing of lead-lag relations among many

variables, understanding the statistical signiﬁcance of these lead-

lag relations and learning which variables are the more important

ones to watch as signals for predicting the market moves. Better

forecasting is the key element for better ﬁnancial decision making,

in the increasing ﬁnancial market volatility and internationalized

capital ﬂows.

Accurate forecasting methods are crucial for portfolio manage-

ment by commercial and investment banks. Assessing expected re-

turns relative to risk presumes that portfolio strategists

understand the distribution of returns. Financial expert can easily

model the inﬂuence of tangible assets to the market value, but

not intangible asset like know-how and trademark. The ﬁnancial

time series models expressed by ﬁnancial theories have been the

basis for forecasting a series of data in the twentieth century.

Studies focusing on forecasting the stock markets have been

mostly preoccupied with forecasting volatilites. There has been

few studies bringing models from other forecasting areas such as

technology forecasting.

To model the market value, one of the best ways is the use of

expert systems with artiﬁcial neural networks (ANN), which do

not contain standard formulas and can easily adapt the changes

of the market. In literature many artiﬁcial neural network models

are evaluated against statistical models for forecasting the market

value. It is observed that in most of the cases ANN models give bet-

ter result than other methods. However, there are very few studies

comparing the ANN models do among themselves, where this

study is ﬁlling a gap.

Objective of this study is to compare performance of most re-

cent ANN models in forecasting time series used in market values.

Autoregressive Conditional Heteroscedasticity (ARCH) model

(Engle, 1982), generalized version of ARCH model Generalized

ARCH (GARCH) model (Bollerslev, 1986), Exponential GARCH

(EGARCH) model (Nelson, 1991) and Dynamic Architecture for

Artiﬁcial Neural Networks (DAN2).

Ghiassi and Saidane (2005) will be analyzed in comparison to

classical Multi-Layer Perceptron (MLP) model. Despite the popular-

ity and implementation of the ANN models in many complex

ﬁnancial markets directly, shortcomings are observed. The noise

that caused by changes in market conditions, it is hard to reﬂect

the market variables directly into the models without any assump-

tions (Roh, 2007). That is why the new models will also be exe-

cuted in hybrid combination with MLP. The analysed models will

be tested on NASDAQ index data for nine months and the methods

will be compared by using Mean Square Error (MSE) and Mean

Absolute Deviation (MAD).

The remaining sections of this paper are organized as follows:

Section 2 gives the background of the related studies; Section 3

introduces the models used in this study and Section 4 provides re-

sults of each model using daily exchange rates of NASDAQ index.

Final section gives the conclusion and recommendations for future

researches.

This study will not only make contribution to the ANN research

but also to the business implementations of market value

calculation.

doi:10.1016/j.eswa.2011.02.068

⇑

Corresponding author.

E-mail address: gkayakutlu@gmail.com (G. Kayakutlu).

Expert Systems with Applications 38 (2011) 10389–10397

Contents lists available at ScienceDirect

Expert Systems with Applications

journal homepage: www.elsevier.com/locate/eswa

2. Background

2.1. Time series forecasting and ANN

The ﬁnancial time series models expressed by ﬁnancial theo-

ries have been the basis for forecasting a series of data in the

twentieth century. Yet, these theories are not directly applicable

to predict the market values which have external impact. The

development of multi layer concept allowed ANN (Artiﬁcial

Neural Networks) to be chosen as a prediction tool besides other

methods. Various models have been used by researchers to fore-

cast market value series by using ANN. A brief literature survey is

given in Table 1.

Gooijer and Hyndman (2006) reviewed the papers about time

series forecasting from 1982 to 2005. It has been prepared for

the silver jubilee volume of international journal of forecasting,

for the 25th birthday of International Institute of Forecasters

(IIF). In this review statistical and simulation methods are analyzed

to include exponential smoothing, ARIMA, seasonality, state space

and structural models, nonlinear models, long memory models,

ARCH-GARCH. Gooijer and Hyndman (2006) compiled the reported

advantages and disadvantages of each methodology and pointed

out the potential future research ﬁelds. They also denoted exis-

tence of many outstanding issues associated with ANN utilisation

and implementation stating when they are likely to outperform

other methods. Last few years researches are focused on improving

the ANN’s prediction performance and developing new artiﬁcial

neural network architecture.

Engle (1982) suggested the Autoregressive Conditional Hetero-

scedasticity (ARCH) model, Bollerslev (1986) generalized the ARCH

model and proposed the Generalized ARCH (GARCH) model for

time series forecasting. By considering the leverage effect limita-

tion of the GARCH model, the Exponential GARCH (EGARCH) model

was proposed by Nelson (1991). Despite the popularity and imple-

mentation of the ANN models in many complex ﬁnancial markets

directly, shortcomings are observed. The noise that caused by

changes in market conditions, it is hard to reﬂect the market vari-

ables directly into the models without any assumptions (Roh,

2007).

Preminger and Franck (2007) used a robust linear autoregres-

sive and a robust neural network model to forecast exchange rates.

Their robust models were better than classical models but still are

not better than Random Walk (RW). Roh (2007) used classical ANN

and EWMA (Exponentially Weighted Moving Average), GARCH and

EGARCH models with ANN. NN-EGARCH model outperforms the

other models with a 100% hit ratio for smaller forecasting period

than 10 days.

Kumar and Ravi (2007) reviews 128 papers about bankruptcy

prediction of banks and ﬁrms. This review shows that ANN meth-

ods outperforms many methods and hybrid systems can combine

the advantages of different methods. Ghiassi, Saidane, and Zimbra

(2005) evaluated ANN, ARIMA and DAN2 (Dynamic Architecture

for Artiﬁcial Neural Networks) using popular time series in litera-

ture. DAN2, is a new NN architecture ﬁrst developed by Ghiassi

and Saidane (2005), clearly outperforms the other methods.

DAN2 is pure feed forward NN architecture and detailed informa-

tion about this architecture will be given in Section 5.

Menezes and Nikolaev (2006) used a new NN architecture and

named it Polynomial Genetic Programming. It is based on Polyno-

mial Neural Network ﬁrst developed by Ivakhnenko (Menezes &

Nikolaev, 2006). This architecture uses polynomials to build an

ANN. Menezes and Nikolaev (2006) uses genetic algorithm to esti-

mate ANN parameters such as starting polynomials, weight esti-

mation etc. This study gives better result for some problems. It is

a new promising architecture but it needs improvement (Menezes

& Nikolaev, 2006).

Zhang and Wan (2007) developed a new ANN architecture Sta-

tistical Fuzzy Interval Neural Network based on Fuzzy Interval

Neural Network. JPY/USD and GBP/USD exchanges rates are pre-

dicted using these methods. These methods are developed to pre-

dict only an interval not a point in time. Hassan, Nath, and Kirley

(2007) used a hybrid model including Hidden Markov Model,

ANN and Genetic Algorithm. They test hybrid model on stock ex-

change rates. Hybrid model is proven to be better than simulation

models.

Yu and Huarng (2008) used bivariate neural networks, bivari-

ate neural network-based fuzzy time series, and bivariate neural

network-based fuzzy time series model with substitutes to apply

neural networks to fuzzy time series forecasting. Bivariate neural

network-based fuzzy time series model with substitutes performs

the best. Zhu, Wang, Xu, and Li (2008) used basic and augmented

neural network models to show trading volume can improve the

prediction performance of neural networks. Leu, Lee, and Jou

(2009) compared radial basis-function neural network (RBFNN),

random walk, and distance-based fuzzy time series models with

daily closing values of TAIEX, and exchange rates NTD/USD,

KRW/USD, CNY/USD, JPY/USD. Results show that RBFNN outper-

formed the random walk model and the artiﬁcial neural network

model in terms of mean square error. Cheng, Chen, and Lin

(2010) used PNN (Prbobabilistic NN), rough sets, and hybrid

model (PNN, Rough Set, C 4.5 Decision Tree) to integrate funda-

mental analysis and technical analysis to build up a trading mod-

el of stock market timing. They report that hybrid model is

helpful to construct a better predictive power trading system

for stock market timing analysis. Chang, Liu, Lin, Fan, and Ng

(2009) used an integrated system (CBDWNN) which combines

dynamic time windows, case based reasoning (CBR), and neural

network (NN). Their CBDWNN model outperformed other com-

paired methods, and very informative and robust for average

investors.

Egrioglu, Aladag, Yolcu, Uslu, and Basaran (2009) introduced a

new method which is based on feed forward artiﬁcial neural

networks to analyze multivariate high order fuzzy time series fore-

casting models. Khashei and Bijari (2010) compaired auto-

regressive integrated moving average (ARIMA), artiﬁcial neural

networks (ANNs), and Zhang’s hybrid model. And Hybrid model

outperforms the other models. Hamzacebi, Akay, and Kutay

(2009) compaired ARIMA and ANN and conclude that direct fore-

cast with ANN is better and noted that before generalizing the con-

clusion other researchs should be done. Majhi, Panda, and Sahoo

(2009) compaired functional link artiﬁcial neural network

(FLANN), cascaded functional link artiﬁcial neural network

(CFLANN),and LMS model and observed that the CFLANN model

performs the best followed by the FLANN and the LMS models.

Liao and Wang (2010) used stochastic time effective neural net-

work model to shows some predictive results on the global stock

indices and their model is showed predictive results. Atsalakis

and Valavanis (2009a) used Adaptive Neuro Fuzzy Inference Sys-

tem (ANFIS) to determine the best stock trend prediction model

and results show that ANFIS clearly demonstrates the potential

of neurofuzzy based modeling for ﬁnancial market prediction.

Chen, Ying, and Pan (2010) also used ANFIS to predict monthly

tourist arrivals. And conclude that ANFIS performs better than

markov and fuzzy models. Bildirici and Ersin (2009) combined

ANNs with ARCH/GARCH, EGARCH, TGARCH, PGARCH, APGARCH.

This combined models better perfomred than ANNs or GARCH

based models. Guresen and Kayakutlu (2008) used hybrid models

like GARCH-DAN2 and EGARCH-DAN2 to forecast Istanbul Stock

Exchange Index (ISE XU100). Yudong and Lenan (2009) used bacte-

rial chemotaxis optimization (BCO), and back propagation neural

network (BPNN) on S&P 500 index and conclude that their hybrid

model (IBCO–BP) model offers less computational complexity, bet-

10390 E. Guresen et al. / Expert Systems with Applications 38 (2011) 10389–10397

剩余8页未读，继续阅读

评论收藏

内容反馈

Quant0xff

粉丝: 1w+
资源: 459

using ANN models in stock market index prediction.pdf

最新资源

using ANN models in stock market index prediction.pdf

Predicting direction of stock price index movement using ANN and SVM.pdf

A fusion model of HMM, ANN and GA for stock market forecasting.pdf

Deep learning for stock market prediction from financial news articles.pdf

Deep Learning Networks for Stock Market Analysis and Prediction

金融股票深度学习论文整理

DeepDTA_Deep Drug-Target Binding Affinity Prediction2018.pdf

Handbook217Plus Reliability Prediction Models.pdf

Stock Price Prediction.ipynb

A Relational Tucker Decomposition for Multi-Relational Link Prediction.pdf

SVM for prediction of futures prices in Indian Stock Market.pdf

Chronic Kidney Disease Prediction.pdf

stock-prediction.zip

Multiplex Graph Neural Network for Tabular Data Prediction.pdf

Stock market prediction performance of neural net- works A literature review.pdf

time-series-prediction.rar_ANN_ann prediction_prediction neural

US-Stock-Market-Prediction-by-LSTM-

kaggle_porto-seguro-safe-driver-prediction.zip

Matlab Module for Stock Market Prediction using Simple NN.zip

porto-seguro-safe-driver-prediction.zip

Stock-Market-Prediction-master_stock_

Aggregating_Time_Series_and_Tabular_Data_in_Deep_Learning_Model_for_University_Students_GPA_Prediction.pdf

track_prediction.cc

论文研究-A novel aging-resilient configurable aging sensor for circuit failure prediction.pdf

rossmann_sales_prediction.ipynb

house-price-prediction.ipynb

最新资源