Scribd
Upload
174 views10 pages

Comparing ANN, NARMAX, TLM

This document outlines a research proposal to compare different methods for predicting stock price movements, including traditional linear models and more advanced artificial neural network models. The research aims to statistically analyze which model best captures expected stock returns. The models to be compared are the Capital Asset Pricing Model, Fama & French Three Factor Model, ARIMA, Artificial Neural Network, and NARMAX. The proposal also describes the data, methodology, models, and accuracy measures that will be used to conduct the comparative analysis and answer the research questions about which models best forecast stock returns.

Uploaded by

Muhammad Abubakr Naeem
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
174 views10 pages

Comparing ANN, NARMAX, TLM

This document outlines a research proposal to compare different methods for predicting stock price movements, including traditional linear models and more advanced artificial neural network models. The research aims to statistically analyze which model best captures expected stock returns. The models to be compared are the Capital Asset Pricing Model, Fama & French Three Factor Model, ARIMA, Artificial Neural Network, and NARMAX. The proposal also describes the data, methodology, models, and accuracy measures that will be used to conduct the comparative analysis and answer the research questions about which models best forecast stock returns.

Uploaded by

Muhammad Abubakr Naeem
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 10

Predicting stock price movements: Comparison of

Artificial Neural Networks, NARMAX, NARIMA


and Traditional Linear Models
Presented by
Muhammad Abubakr Naeem

Research Objective
The objective proposed in this research proposal is to statistically
compare different methods able to construct expected stock return
estimates. To determine which model performs best in capturing
expected stock returns, it is proposed to conduct a comparative
statistical analysis.

Capital Asset Pricing Model (CAPM)


Fama & French Three factor Model (FF3)
Autoregressive Integrated Moving Average Model (ARIMA)
Artificial Neural Network Model (ANN)
Non-linear Autoregressive Moving Average Model with Exogenous Inputs
(NARMAX)

Research Questions
Which models are best in capturing expected stock returns?
Can more sophisticated statistical models, such as the neural network
forecasting models, capture expected stock returns better than the
equilibrium models?

Data & Methodology


The data would consist of daily closing prices of stocks and
annual book value and common shares outstanding of the
selected firms.
The softwares used for the purpose of empirical analysis
are:
MS Excel
STATA
MATLAB

Linear Models
The
accuracy of forecasting of an autoregressive AR (p) model can be improved with the addition of
a moving-average MA (q) component (Watts and Leftwich 1977; Brown and Rozeff 1979; Griffin
1977).

In the above equations,


= Stock returns, = Constant, = Autoregressive component,
= RPM = (, = Small minus Big (SMB),
= High minus Low (HML), = Moving Average component,

Artificial Neural Network Models


Based
on the generic three layer feed forward neural network model, the three artificial neural network
models can be formally written as:

The above models represent the ANN versions of the ARIMA, dynamic CAPM and dynamic FF-three
factor model.

NARMAX ()

The
nonlinear autoregressive moving average model with exogenous inputs is defined as

Where;
y, e and x are output, noise and external input of the system model respectively, , and are the
maximum lags in the output, noise and input respectively, and F is an unknown smooth function.
Several special cases of the general NARMAX ( ) model are frequently seen, which are represented as;
NAR () Model
NARMA (, ) Model
NARX (, ) Model

Forecast Accuracy
For
accessing the accuracy of forecast, the following measures will be used. (Calen et
al. (1996);
Zhang et al. (2004).

Where, N is the number of total observations,


n is the number of observations used for the purpose of estimation,
is the actual value of stock returns and
is the forecasted value of stock returns.

Thank You

You might also like