Modeling Volatility and Forecasting of Stock Price: A Case Study on Two Private Commercial Banks in Bangladesh Md. Kamruzzaman Lecturer Department of Statistics, Jagannath University Saifur Rahman Shohel Graduate Student School of Business, Uttara University Md. Mohsan Khudri Assistant Professor School of Business, Uttara University
21
Embed
Modeling Volatility and Forecasting of Stock Price_A Case Study on Two Private Commercial Banks in Bangladesh
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Modeling Volatility and Forecasting of Stock Price: A Case Study on Two Private Commercial Banks in Bangladesh
Md. KamruzzamanLecturer
Department of Statistics, Jagannath University
Saifur Rahman ShohelGraduate Student
School of Business, Uttara University
Md. Mohsan KhudriAssistant Professor
School of Business, Uttara University
Thanks to the Organizers of this conference
PHASE 1: INTRODUCTION
PHASE 2: DATA & VARIABLES
PHASE 3: METHODOLOGY
PHASE 4: ANALYSIS & DISCUSSION
PHASE 5: CONCLUSION
PRESENTATION OUTLINE
OVERVIEW
Forecasting on stock price is a common practice around the world.
We have considered Autoregressive Integrated Moving Average (ARIMA) model
to forecast month ended stock price.
Banks play major rules in economy
of BangladeshTotal 66 banks in
Bangladesh.
More than 6000 branches
Introduction
Introduction
To find out the appropriate forecasting model for the month ended stock market
price of the selected banks.
To forecast the month ended stock price for next 24 months (Jan-2014 to Dec-2015).
To see the forecasting performance of the selected models.
OBJECTIVES OF THE STUDY
Introduction
LITERATURE REVIEW
In “Time series forecasting using a hybrid ARIMA and neural network model”
(2003), Zhang said that Autoregressive integrated moving average (ARIMA) is one
of the popular linear models in time series forecasting during the past three
decades.
According to Pai & Lin in “A hybrid ARIMA and support vector machines model in
stock price forecasting”(2005), The real data sets of stock prices were used to
examine the forecasting accuracy of the proposed model. The results of
computational tests are very promising.
According to Al-Zeaud in “Modeling volatility using ARIMA model in European
journal of Economics” (2011); The study presents the Box-Jenkins model as one of
the forecasting techniques, which we can use, in the financial time series.
AND SO ON………………………..
Data & Variables
VARIABLES TIME PERIOD
Month ended Stock price of National Credit & Commerce Bank Ltd. (NCC) JAN-2001 to DEC-2013
Month ended Stock price of Mutual Trust Bank Ltd. (MTB) JUL-2003 to DEC-2013
DATA SOURCE:Dhaka Stock Exchange (DSE) library of Bangladesh.
DATA & VARIABLES
Bank selection procedure:
Simple random sampling
Out of 66 banks, 2 banks are selected
Package: Statistical package for social science (SPSS)
Methodology
ARIMA Methodology:Time Series Data
Checking stationarity Augmented Dickey-Fuller Test (ADF Test)
Obtaining stationarity Differencing method
FORECASTING METHODOLOGY
ARIMA Model is given below:
qtqtttptd
ptd
td
td eeeec ....... 22112211
Methodology
FORMULATION OF ARIMA MODEL
Order of the autoregressive
part
Degree of difference involved
Order of the moving average
AR MAI
d qp
ARIMA (p,d,q)
Methodology
.
The AIC is given by:
Where , L = maximum likelihoodm = is the number of terms estimated in the model.
SELECTION OF BEST ARIMA MODEL
The model having the minimum AIC value will be treated as the best model.
Packages used: Statistical package for social science (SPSS) R
2m2logLAIC
Akaike’s Information Criterion (AIC)
Measure of forecast error
Analysis & Discussion
CHECKING STATIONARITY OF DATA
Figure: Time series plot of month ended stock price of National Credit & Commerce Bank Ltd.
Figure: Time series plot of month ended stock price of Mutual Trust Bank Ltd.
Figure: Time series plot of observed data
Analysis & Discussion
Null-hypothesis = Data is Non-Stationary,Alternative Hypothesis = Data is Stationary,Significance level, = 0.05α
Results of Augmented Dickey-Fuller (ADF) Test:
CHECKING STATIONARITY OF DATA
Since the p>α for NCC Bank, so we can’t reject Null-hypothesis. So data is non-stationary.
Since the p>α for Mutual Trust Bank, so we can’t reject Null-hypothesis. So data is non-stationary.
Variable p-value Note Data type
NCC 0.5981 p>α NON-STATIONARY
MTB 0.4488 p>α NON-STATIONARY
Analysis & Discussion
OBTAINING STATIONARITY USING DIFFERENCING METHOD
We can difference the data to obtain stationarity. That is, given the series , we create the new series , using the equation below:
1 ttt YYZ
tY tZ
Analysis & Discussion
Figure: Time series plot of first difference month ended stock price of National Credit & Commerce Bank Ltd.
Figure: Time series plot of first difference month ended stock price of Mutual Trust Bank Ltd.
Figure: Time series plot of first differenced data
OBTAINING STATIONARITY USING DIFFERENCING METHOD
Analysis & Discussion
Null-hypothesis = Data is Non-Stationary,Alternative Hypothesis = Data is Stationary,Significance level, = 0.05α
Results of Augmented Dickey-Fuller Test after obtaining stationarity:
Since p<α for NCC Bank, so we can reject Null-hypothesis. So data is now stationary.
Since p<α for Mutual Trust Bank, so we can reject Null-hypothesis. So data is now stationary.