3. INTRODUCTION
● The stock market is considered to be one of the most highly complex financial systems
which consist of various components or stocks, the price of which fluctuates greatly with
respect to time
● All the stock market investors aim to maximize the returns over their investments and
minimize the risks associated
● Stock markets being highly sensitive and susceptible to quick changes, the main aim of
stock-trend prediction is to develop new innovative approaches to foresee the stocks that
result in high profits.
● This research tries to analyze the time series data of the Indian stock market and build a
statistical model that could efficiently predict the future stocks.
TKMCE
2
21/10/2020
5. Literature Review
TKMCE
4
Title Author synopsis
1 Stock index forecasting based
on a hybrid model
J.-J. Wang, J.-Z. Wang,
Z.-G. Zhang, and S.-P.
Guo
Using Feed forward back
propagation neural network to
forecast stock prices
2
Support Vector Machine With
Adaptive Parameters in
Financial Time Series
Forecasting
L. J. Cao and Francis E.
H. Tay
The variability in performance of
SVM with respect to the free
parameters is investigated
experimentally. Adaptive parameters
are then proposed by incorporating
the non stationarity of financial time
series into SVM
3
An introductory study on time
series modeling and forecasting.
R. Adhikari and R. K.
Agrawal.
It have described three important
classes of time series models, ie.the
stochastic, neural networks and
SVM based models, together with
their inherent forecasting strengths
and weaknesses.
6. Time Series Forecasting
● Time series forecasting is a technique for the prediction of events through a sequence of time .
● Time series data refers to an ordered sequence or a set of data points that a variable takes at equal
time intervals
● In time series ,time acts as an independent variable to estimate dependent variable
Source:https://www.researchgate.net/figure/Figure-No2-Time-series-graph-Timber-production-from-year-2000-
2005-in-tonnes_fig1_257985747TKMCE
5
21/10/2020fig 3
7. Time Series Components
● The components by which time series is composed are called the components of time series
❖ Trend component (T)
-It is the general tendency of data to grow or decline over a long
period of time
❖ Seasonal component (S)
-The Component responsible for regular rise or fall during a period
not more than 1 year
❖ Cyclic component (C)
-These are the recurrent variations in time series
❖ Irregular component (I)
- These component don’t repeat in a definite pattern
TKMCE
6
21/10/2020
8. Forecasting Model -ARIMA
● ARIMA stands for Auto Regressive Integrated Moving Average
● It is an integrated model of Auto Regressive(AR) and Moving Average(MV)
● In this model past values of the time series alone used to predict the future values
● The common form of an ARIMA model
● This model is called as “ARIMA (p ,d , q) model
where,
p - order of auto-regressive part
d - degree of differencing
q – order of the moving average part
TKMCE
7
21-10-2020
10. Methodology
1. Identify
● In order to design ARIMA model, the primary time series has to be Stationary one
● If the series is non-stationary, then the series has to be differenced so as to make it stationary
2. Estimation
● An important step while selecting the model is the determination of ideal parameters for the
model
● Plotting the ACF and PACF against the consecutive time lags for the series is a simple
approach to choose the parameters of the model
● The general form of ACF is as:
Covariance (Xt, Xt − h)/ Variance(Xt)
● The common form is of PACF is as:
Covariance (y, X3|X1, X2)/ variance(y|X1|X2)variance(X3|X1, X2)TKMCE
9
21/10/2020
12. 3. Model Selection
● We use the ‘‘ARIMA (0, 1, 0) model’’ for predicting the next values in the time series.
● We use the auto.arima () function in R to get the results.
● Auto.arima() function chooses the best parameters of ARIMA(p,d,q).
● Test time series data from Nifty
auto.arima (lnstock_Nifty, ic=‘‘aic’’, trace = TRUE)
‘‘ARIMA(2,1,2)’’ : −166.3623
‘‘ARIMA(0,1,0)’’ : −168.2363
‘‘ARIMA(1,1,0)’’ : −166.8252
‘‘ARIMA(0,1,1)’’ : −166.8558
‘‘ARIMA(1,1,1)’’ : −167.7504
‘‘Best model: ARIMA(0,1,0)’’
TKMCE
11
21/10/2020
16. ● In this ARIMA model, used different p,q,d values to get best result
● Model ARIMA(0,1,0) got the best result
● comparison of the predicted series with the actual series shows roughly a deviation of 5% mean
percentage error for both Nifty and Sensex on average
RESULTS & DISCUSSIONS
TKMCE
15
17. CONCLUSION
● This research was taken to investigate the utility of Time Series Forecasting in Indian stock
market
● Here we tries to build an efficient ARIMA model to predict the Indian stock market volatility
● The publically available time series data of Indian stock market has been used for this study
● The proposed Time Series Forecasting model for the stock market prediction got a good
result with a roughly deviation of 5% mean percentage error.
● Various tests can be used for the validation of the predicted time series. However, in this
study we have used the ‘‘ADF test and the L-jung box tests’’ for purpose of validation
● It would be useful when we investing in stock market for a long period
TKMCE
16
21/10/2020
18. ACKNOWLEDGEMENT
I would like to express my deep sense of gratitude to my guide yyyyy for providing
invaluable guidance, comments, and suggestions throughout the course of seminar.
I am very much thankful to all the faculties from the ECE department especially Prof.
xxxxx (seminar coordinator) for providing me an opportunity to present my seminar.
Last but not the least I want to thank my friends for attending the seminar and listening me
from the last few minutes.
17
TKMCE 21-10-2020
19. REFERENCES
1. G. González’Rivera and T. H. Lee, ‘‘Nonlinear time series in financial forecasting,’’ in Encyclopedia of
Complexity and Systems Science. New York, NY, USA: Springer, 2009. .
2. P.-F. Pai and C.-S. Lin, ‘‘A hybrid ARIMA and support vector machines model in stock price
forecasting,’’ Omega, vol. 33, pp. 497–505, Dec. 2005
3. An Introduction to Indian Stock Market. Accessed: Jul. 2018. [Online]. Available:
https://www.investopedia.com/articles/stocks/09/indian-stockmarket.asp
4. G. P. Zhang, ‘‘A neural network ensemble method with jittered training data for time series forecasting,’’
Inf. Sci., vol. 177, no. 23, pp. 5329–5346, 2007.
5. S. Green, ‘‘Time series analysis of stock prices using the box-Jenkins approach,’’ Tech. Rep., 2011
6. J. Pati, B. Kumar, D. Manjhi, and K. K. Shukla, ‘‘A comparison among ARIMA, BP-NN, and MOGA-NN
for software clone evolution prediction,’’ IEEE Access, vol. 5, pp. 11841–11851, 2017.
7. S. M. Idrees, M. A. Alam, and P. Agarwal, ‘‘A study of big data and its challenges,’’ Int. J. Inf. Technol.,
pp. 1–6, 2018.
8. NIFTY 50 (NSEI)/S&P BSE SENSEX (BSESN). Accessed: Jul. 15, 2018. [Online]. Available:
https://in.finance.yahoo.com
9. T. G. Andersen, T. Bollerslev, F. X. Diebold, and P. Labys, ‘‘Modeling and forecasting realized
volatility,’’ Econometrica, vol. 71, no. 2, pp. 579–625, 2003
TKMCECE 01-10-2020
18
21/10/2020