Predicting Market Reactions Using Articial Intelligence
Ruchira Ray and Prakhar Khandelwal
Abstract — The stock market is a crucial component of any
economy in the world. It is a way for companies to gain capital
for its day to day functions. It enables stock brokers to trade
securities, bonds and equities in a market. Once a share is listed
on the stock market, it can be bought and sold by traders,
investors or the general public. Recently, a lot of work has
been done to predict the movement of the market. Forecasting
the movement of the stock market is gaining momentum among
various researchers, investing communities and enthusiasts as it
provides better guidance with respect to investing. Predictability
is one of the major factors which, the protability of trading
in stock and investing is dependent on. The prots earned by
investment and trading in the stock market depend on the
predictability of the stock, to a large extent. If any system
is developed which can consistently predict the trends of the
dynamic stock market, would make the owner of the system
wealthy. Moreover, the predicted trends of the market will help
the regulators to make corrective measures to stabilize the
market. Many expert practitioners and researchers have put
forward several models using various technical, fundamental
and analytical techniques to give a more or less prediction on
the stock market pattern.
Stock market trading is a renowned way of earning money
in a short span of time. Investors nd it very difcult to
predict the market prices. The prices not only depend on
nancial events and the companys performance but also on
the psychology of people investing in the market. In this
survey paper, we have tried to analyze the work done in
past few years in the eld of stock market prediction. We
have examined the techniques used and results achieved
by the fellow researchers. We have also tried to study the
shortcomings of their work in this paper.
II. RELATED WORK
The stock market is a very rewarding eld and has
generated interest from a lot of researchers. A considerably
large number of intelligent people have made models to
predict the outcome of the market. Machine learning and
articial intelligence enable us to make intelligent models to
forecast market movements. In this survey paper, we have
tried to analyze the work done by fellow researchers in the
eld of stock market prediction.
A. Stock market prediction using machine learning tech-
The paper on Stock market prediction using machine
learning techniques by Mehak Usmani, Syed Hasan Adil,
Kamran Raza, and Syed Saad Azhar Ali predicts the per-
formance of Karachi Stock Exchange (KSE) on day closing.
Their methodology uses single-layer perceptron, multi-layer perceptron, radial basis function and support vector ma-
chines. The factors used as input attributes are market history,
news, general public mood, commodity price, interest rate,
and foreign exchange. As per their simulation, the multi-layer
perceptron achieved an accuracy of 77 percent. Further, it
was realized that petrol price was the most related attribute,
whereas foreign exchange had no impact on stock prices.
B. Intelligent Stock Data Prediction using Predictive Data
In the year 2016, the paper on Intelligent Stock Data Pre-
diction using Predictive Data Mining Techniques by Dr. Anju
Bala and Pankaj Kumar shed light on predictive data mining
algorithm for market prediction. Their dataset consists of 21
features followed by 57772 data entries and target at the 22nd
position. They classied the data based on the decision tree,
linear model and random forest. Their analysis shows that
the accuracy of the random forest is higher than the linear
model and decision tree. As an extension to their work, we
can merge various models to gain more accuracy.
C. Short Term Stock Price Prediction using Deep Learning
The use of deep learning techniques to predict short-term
stock market prices was shown in the paper presented by
Kaustubh Khare, Omkar Darekar, Prafull Gupta and Dr. V.
Z. Attar. The paper titled Short Term Stock Price Prediction
Using Deep Learning Techniques attempts to forecast the
transient future prices of the stock under consideration.
Their methodology makes use of Feed Forward Multilayer
Perceptron and Long Short-Term Memory model to analyze
the market. The dataset consists of a minute to a minute
stock price of 10 shares listed on NSE, over the period of
one year. Each stock consists of 85000 to 90000 points. The
study shows that MLPs have performed better at predicting
short-term stock market price than LSTM. To extend further,
we can use the Indian Stock Market data and build a platform
to conduct trades based on data obtained.
D. Predicting Market Prices using Deep Learning Tech-
One more research was conducted on the effect of deep
learning techniques on stock price prediction. The research
conducted by Nishanth C, Dr. V K Gopal, Vinayakumar R,
Lakshmi Nambiar, and Dileep G Menon, titled Predicting
Market Prices Using Deep Learning Techniques uses a
model-independent approach. As opposed to previous re-
searchers, the authors have used Recurrent Neural Network,
Long Short-Term Memory and Gated Recurrent Unit in their
model. Tata Motors and Syndicate Bank data was used to
train and predict the data in their respective elds. The
model obtained the best result under the LSTM approach.
To capture irregular changes in the stock market, we can
use the Convolution Neural Network as an extension to the
E. Analyzing stock price chnages using event-related Twitter
Satyabrata Aich, Hee-Cheol Kim, Mangal Sain, and Bijay
Bhaskar Deo conducted a research to analyze the sentiments
of people, and its effect on stock market using data from
the Twitter platform. The paper titled, Analyzing stock price
changes using event-related Twitter feeds, outlines the stock
prices changes based on the event-related day wise tweet
sentiment score. They used Tweepy, a python library to
access the Twitter API. The tweets were ltered based on
the keyword Samsung . A collection of 200000 tweets were
made on the launch of galaxy note 7. Sentiment analysis
and keyword analysis was carried out on the given data. They
were able to nd some positive correlation between data sets.
In the future, more tweets with different techniques can be
analyzed for a better performance.
F. Proposed System for Estimating Intrinsic Value of Stock
Using Monte Carlo Simulation
Sehba Shahabuddin Siddiqui and Vandana A. Patil con-
ducted a research to propose a system for estimating intrinsic
value of stock using Monte Carlo Stimulation. Based on
the drawbacks of the existing system, they tried to nd
intrinsic values and displayed the results by visualisation.
Their shortcoming was the low delta factor which can be
further improved. In future, values of past and prospective
investments can be calibrated.
G. Survey of Stock Market Prediction using Machine Learn-
Ashish Sharma, Dinesh Bhuriya and Upendra Singh anal-
ysed the stock market values to predict future values using
regression. What they faced as shortcoming was the lack
of t in the least squares approach. In future, multiple
regression approach can be improved by using more number
H. A Review of Stock Market Prediction with Articial
Neural Network (ANN)
Chang Sim Vui, Gan Kim Soon, Chin Kim On, and Rayner
Alfred, Patricia Anthony gave an overview or survey on
market prediction using articial neural networks where it
was found that the conventional ANN has lower accuracy.
This paper can be referred to as an introductory material.
I. Stock Market Prediction Using an Improved Training
Algorithm of Neural Network
Mustain Billah, Sajjad Waheed and Abu Hanifa used
Adaptive Neuro Fuzzy Inference Systems(ANFIS) for better results than Neural Networks for stock prediction. In this pa-
per, they used Levenberg Marquard(LM) training algorithm
of ANN, which can predict the day end closing stock price
with less memory and time. This technique can predict stock
price with 53 percent less error than Adaptive Neuro Fuzzy
Inference System and Traditional LM algorithm. It requires
30 percent less time, 54 percent less memory than traditional
LM and 47 percent less time, 59 percent less memory than
J. Stock Prediction and Analysis Using Intermittent Training
Data With Articial Neural Networks
N. Srinivasan and C. Lakshmi did a research, where based
on Multiple layer neural networks they have predicted the
stock market values. The training vector along with training
data had input layers towards the output layers.This tech-
nique foretells the prices of stocks and shares of companies
indexed under National Stock Exchange.
We would like to express our deep gratitude to Dr.
Baranidharan Balakrishnan our project guide, for her patient
guidance, enthusiastic encouragement and useful critiques
on this project work. She has helped on every stage of the
project. His willingness to give us so much of his time is
very much appreciated.
We would also like to thank Mrs. S.S.Saranya and Mr.
Arul Prakash for her advice and assistance in keeping our
progress on schedule. We are really grateful for her help
in nding the papers and also for her valuable suggestions
throughout the semester. Our grateful thanks are also extended to our HOD Dr. B.
Amutha for the encouragement and providing all the facilities
in the department.
RE F E R E N C E S
1 Ian Leifer, Lev Leifer, Small Business Valuation with Use of Cash Flow Stochastic Modeling, in Second International Symposium on
Stochastic Models in Reliability Engineering, Life Science and Oper-
ations Management, pp. 511-516,IEEE,2016.
2 Sehba Shahabuddin Siddiqui, Vandana A. Patil, Proposed system for estimating intrinsic value of value of stock using monte carlo
simulation, in International Conference on Intelligent Computing and
Control Systems, ICICCS 2017
3 Ashish Sharma, Dinesh Bhuriya, Upendra Singh, Survey of stock
market prediction using Machine learning approach, International Con-
ference on Electronics, Communication and Aerospace Technology,
4 M.P. Naeini, H. Taremian, and H.B. Hashemi, Stock market value pre- diction using neural networks, International Conference on Computer
Information Systems and Industrial Management Applications (CISIM
2010), pp. 132-136, 2010.
5 M. Mehrara, A. Moeini, M. Ahrari, and A. Ghafari, Using Technical Analysis with Neural Network for Prediction Stock Price Index in
Tehran Stock Exchange, Middle Eastern Finance and Economics, vol.
6(6), pp. 50-61, 2010.
6 R. Dase, and D. Pawar, Application of articial neural network for stock market predictions: a review of literature, International Journal
of Machine Intelligence, vol. 2(2), pp. 1417, 2010. International
Conference on Computer and Network Technology (ICCNT), pp. 377-