Scribd
Upload
27 views2 pages

IEEE Template

This review discusses the use of Artificial Neural Networks (ANNs) in stock market index prediction, focusing on three models: Multilayer Perceptron (MLP), Dynamic Artificial Neural Network (DAN2), and a hybrid GARCH-MLP model. While MLP effectively captures historical trends, it struggles with external influences, and DAN2 faces limitations in real-time forecasting due to its need for retraining. The findings emphasize the importance of model selection and the need for continuous innovation in ANN methodologies for improved financial forecasting.

Uploaded by

hktom20
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
27 views2 pages

IEEE Template

This review discusses the use of Artificial Neural Networks (ANNs) in stock market index prediction, focusing on three models: Multilayer Perceptron (MLP), Dynamic Artificial Neural Network (DAN2), and a hybrid GARCH-MLP model. While MLP effectively captures historical trends, it struggles with external influences, and DAN2 faces limitations in real-time forecasting due to its need for retraining. The findings emphasize the importance of model selection and the need for continuous innovation in ANN methodologies for improved financial forecasting.

Uploaded by

hktom20
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 1

Review on Using Artificial Neural Network


Models in Stock Market Index Prediction
(November 2024)
Cheung Chi Hei

Since the dynamic and unpredictable nature of financial solve the heteroskedasticity in financial time series and to
markets, forecasting the stock market is constantly challenging. capture the volatility clustering and leverage effects. After
Artificial neural networks (ANNs), a tool used to handle integration, the hybrid model is expected to have a higher
financial forecasting accurately and effectively, have gradually accuracy in its prediction. However, the paper mentions that the
been used in the financial markets. This review will briefly list hybrid model did not outperform the MLP model since the
out the Pros and Cons of three ANN models: Multilayer performance of the model was often encumbered by the noise
Perceptron (MLP), Dynamic Artificial Neural Network introduced by the GARCH inputs . This demonstrates that more
(DAN2), and a hybrid model incorporating GARCH elements advanced methods are needed to effectively Integrate statistical
according to the evaluation of Guresen, Kayakutlu, and Daim and neural components.
(2011)
In conclusion, the above findings discuss the effectiveness of
The MLP model performs well in handling nonlinear different ANNs, the importance of model selection, and the
patterns in financial time series. Using interconnected neuron need for continuous innovation in ANN.
layers with nonlinear activation functions and training through
a backpropagation algorithm, MLP can learn complex
input-output relationships and make accurate short-term
predictions. The paper highlights that the MLP model performs
well when applied to financial datasets by capturing historical
trends and adapting to market fluctuations. However, the
reliability of historical trends also causes its downside. MLP
only consists of historical trends but ignores external influences,
such as geopolitical events or economic policy, which might
significantly affect the market trend. These limitations would
negatively affect MLP's prediction of new and long-trend
markets.

The study also includes the DAN2 model, a new ANN


architecture developed by Ghiassi and Saidane. The difference
between DAN2 and traditional ANNs is that DAN2 is a
dynamic architecture that iteratively adapts its structure to
minimize prediction errors. However, the paper also identifies
some issues with the DAN2 model, which limits the
performance of DAN2, such as regression-based initialization
and the inability to adapt new data without retraining. The
real-time financial forecasting performance of DAN2 is
therefore being questioned due to the above limitations. In
order to release the model's potential for real-time financial
forecasting, the model will need further improvements.

The hybrid GARCH-MLP model is integrated by including


GARCH models into ANNs. GARCH models are designed to
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 2

References
[1] Guresen, E., Kayakutlu, G., & Daim, T. U. (2011). Using artificial neural
network models in stock market index prediction. Expert Systems with
Applications, 38(8), 10389–10397.W.-K. Chen, Linear Networks and
Systems (Book style).

You might also like