Stock Market Prediction Using Machine

Learning
Domain – Machine Learning

Submitted by :- 1) M.ANIL (U17CN030)

2) K.VISWANATH REDDY (U17CN002)
 Abstract

 Predicting stock market can be risky as it has trends and seasonality, using
conventional analysis, it may not predict close to actual values.
 In order to predict with high accuracy latest algorithms of machine learning and
time series analysis is needed.
 To achieve this Autoregressive Integrated Moving Average, Exponential
smoothing and support vector machine methods will be used to predict future.
 Stock brokers and Intra traders will have a better hand on stocks with this model.
 This will also help an individual to increase his total assets for secured future.
 Introduction

• A correct prediction of stocks can lead to huge profits for the seller and the broker. Frequently, it is
brought out that prediction is chaotic rather than random, which means it can be predicted by
carefully analyzing the history of respective stock market.
• Machine learning is an efficient way to represent such processes. It predicts a market value close to
the tangible value, thereby increasing the accuracy.
• Introduction of machine learning to the area of stock prediction has appealed to many researches
because of its efficient and accurate measurements .
• The vital part of machine learning is the dataset used. The dataset should be as concrete as possible
because a little change in the data can perpetuate massive changes in the outcome .
 Existing Systems and
Disadvantages

Many apps and websites try to forecast the stock prices for a certain time period.
The technique used by them is exponential smoothing which at times could not
exactly track the stock prices trend. And the type of visualization used is candle
sticks which can be overwhelming for a beginner and it takes more time
comprehend, contradicting gestalt principles of data visualizations.
 LITERATURE SURVEY

• M. Usmani, S. H. Adil, K. Raza and S. S. A. Ali, "Stock market prediction

using machine learningtechniques,"
• This paper faces the Stock market prediction problem using traditional machine
learning models such as support vector machines.
• They have considered different firms to analyze and predict the prices.
• Data has been categorized year to year and different amount of time period was
considered to better estimate the price with close to the ground truth.
 LITERATURE SURVEY

• Stock Market Prediction Using Machine Learning Techniques

• Predicting stock market prices is crucial subject at the present economy. Hence, the tendency of
researchers towards new opportunities to predict the stock market has been increased. Researchers
have found that, historical stock data and Search Engine Queries, social mood from user generated
content in sources like Twitter, Web News has a predictive relationship to the future stock prices. Lack
of information such as social mood was there in past studies and in this research, we discuss an
effective method to analyze multiple information sources to fill the information gap and predict an
accurate future value. For this, LSTM - RNN models were employed to analyze sperate sources and
Ensembled method with Weighted Average and Differential Evolution technique were used for more
accurate prediction of the stock prices. And highly accurate predictions were made to one-day, seven-
days, 15-days and 30 days for the future. So that investors could gain an insight into what they are
inventing for and the companies to track how well they will perform in the stock market.
 LITERATURE SURVEY

• Prediction of Stock Market performance by using machine learning techniques

•  The model was built to predict performance of KSE-100 index. The prediction model predicts
market as positive or negative with the help of different attributes. These factors include price
fluctuation of fuel, commodity, foreign exchange, interest rate, general public sentiment, related
NEWS and Auto-Regressive Integrated Moving Average (ARIMA) and Simple Moving Average
(SMA) predicted values with help of historical data of the market. The techniques used for
prediction include four different versions of Artificial Neural Network (ANN) including Single
Layer Perceptron (SLP), Multi-layer Perceptron (MLP), Radial Basis Function (RBF) and Deep
Belief Network (DBN). Other techniques include Support Vector Machine (SVM), Decision Tree
and Naive Bayes. All these techniques were compared to find the best predicting model. The
results showed that MLP performed best and predicted the market with accuracy of 77%. 
 LITERATURE SURVEY

• Stock market direction prediction using deep neural networks

• In this study, the daily movement directions of three frequently traded stocks (GARAN, THYAO
and ISCTR) in Borsa Istanbul were predicted using deep neural networks. Technical indicators
obtained from individual stock prices and dollar-gold prices were used as features in the prediction.
Class labels indicating the movement direction were found using daily close prices of the stocks
and they were aligned with the feature vectors. In order to perform the prediction process, the type
of deep neural network, Convolutional Neural Network, was trained and the performance of the
classification was evaluated by the accuracy and F-measure metrics.  the movement directions in
GARAN, THYAO and ISCTR stocks were predicted with the accuracy rates of 0.61, 0.578 and
0.574 respectively. Compared to using the price based features only, the use of dollar-gold features
improved the classification performance.
 LITERATURE SURVEY

• Stock market prediction using an improved training algorithm of neural network

•  In this paper, an improved Levenberg Marquardt(LM) training algorithm of artificial neural
network has been proposed. Improved Levenberg Marquardt algorithm of neural network can
predict the possible day-end closing stock price with less memory and time needed, provided
previous historical stock market data of Dhaka Stock Exchange such as opening price, highest
price, lowest price, total share traded. Morever, improved LM algorithm can predict day-end stock
price with 53% less error than ANFIS and traditional LM algorithm. It also requires 30% less time,
54% less memory than traditional LM and 47% less time, 59% less memory than ANFIS.
 LITERATURE SURVEY

• Regression techniques for the prediction of stock price trend

• This paper examines the theory and practice of regression techniques for prediction of stock price
trend by using a transformed data set in ordinal data format. The stock price movement in Bursa
Malaysia is used as our research setting. The data sources are corporate annual reports which
included balance sheet, income statement and cash flow statement. The variables included in the
data set were formed based on stock market trading fundamental analysis approach. Classifiers in
WEKA were used as algorithms to produce the outcomes. This study showed that the outcomes of
regression techniques can be improved for the prediction of stock price trend by using a dataset in
standardized ordinal data format.
 LITERATURE SURVEY

• Stock index prediction using regression and neural network models under non normal
conditions
• Nonparametric Linear Regression and Artificial Neural Network models have been developed
based on different perspectives and assumptions. In this paper a survey is made to compare the
predictive performances of the nonparametric models of closing prices of Stock Index data, where
the data is non normal. Comparative studies with the existing statistical prediction models indicate
that the proposed neural network model is very promising and can be implemented into real time
trading system for stock price prediction.
 LITERATURE SURVEY

• Stock transaction prediction modeling and analysis based on LSTM

• LSTM (Term Memory Long-Short) is a kind of time recurrent neural network, which is suitable for
processing and predicting the important events of interval and long delay in time series. Based on
temporal characteristics of stock and LSTM neural network algorithm, this paper uses the LSTM
recurrent neural networks to filter, extract feature value and analyze the stock data, and set up the
the prediction model of the corresponding stock transaction.
 LITERATURE SURVEY

• Applying long short term momory neural networks for predicting stock closing price
• They propose using long short term memory (LSTM) and stock basic trading data to realize the
stock prediction model. For training the model, we utilize some optimization strategies, such as
adaptive moment estimation (Adam) and glorot uniform initialization. They present a case study
based on Standard & Poor's (S&P500) and NASDAQ. Quantities of comparison experiments were
performed to evaluate this model. At last they analyze the performance of different models with a
series of evaluation criteria. Stock market prediction has garnered significant interest among
investment and researchers. However, accurate prediction of stock market is an extremely
challenging task. Hopefully, based on the case study, they show that our forecasting system gives
slightly higher prediction accuracy for the stock closing price of next day, which outperforms the
comparison models.
 LITERATURE SURVEY

• Stock Closing Price Prediction using Machine Learning SVM Model

• This research was prepared to predict the closing price of the stock in the Stock Exchange of
Thailand (SET). We are using the Multi-Layer Perceptron model, Support Vector Machine model,
and Partial Least Square Classifier to predict the closing price of the stock. In the present, people
have more knowledge and understanding of investing in the stock market then the Thai stock
market has grown significantly. From the statistical data, we can find the movement of stock prices
in that stock market move in a cycle. Form this point; we have the idea that if we can predict the
stock price nearby real price. We can be investing at the right time and help investors to reduce
investment risks. The experimental result shows that Partial Least Square is the best algorithm of
the three algorithms to predict the stock closing price.
 LITERATURE SURVEY

• Evaluation of bidirectional LSTM for short-and longterm stock market prediction

• In this paper, they aim at evaluating and comparing LSTM deep learning architectures for short-
and long-term prediction of financial time series. This problem is often considered as one of the
most challenging real-world applications for time-series prediction. Unlike traditional recurrent
neural networks, LSTM supports time steps of arbitrary sizes and without the vanishing gradient
problem. We consider both bidirectional and stacked LSTM predictive models in our experiments
and also benchmark them with shallow neural networks and simple forms of LSTM networks. The
evaluations are conducted using a publicly available dataset for stock market closing prices.
 LITERATURE SURVEY

• Empirical examination of Indian markets

• The purpose of this paper was to investigate the market integration among the four selected BRIC
economies namely Brazil, Russia, India, and China from January 2008 to August 2015. Johansen
cointegration test was used to study the long term relationship between the four stock markets.
Further, to ascertain the short term association, vector error correction model and impulse response
function were used. The results showed that there existed one long run cointegrating relationship
between the four stock markets under study. Although there was no long run causality among the
four stock markets, but there existed short term causality running from Russian, Chinese, and
Brazilan stock markets to the Indian stock market.
 LITERATURE SURVEY

• How Good is the Transformation-Based Approach to Estimate Value at Risk? Simulation and
Empirical Results
• This paper aims to estimate Value at Risk (VaR) of Tadawul All Shares Index of Saudi Stock
Market (TASI) over the period January 2004 – December 2017. It applies the following methods,
empirical quartile, historical simulation (HS), and percentile, parametric via delta normal,
GARCH, IGARCH, Monte Carlo simulation and bootstrapping simulation. It uses 5% and 1%
critical value under Normal distribution. Back- testing based on likelihood ratio LR accepted
empirical quartile at both five and one percent, while accepting delta normal, historical simulation,
percentile, IGARCH, and Monte Carlo at one percent. The worst loss obtained is approximately
4%.
 LITERATURE SURVEY

• How Good are Bank Loan Ratings in India for Listed and Unlisted Firms
• Basel bank capital norms increase the role of credit ratings for bank lending I A majority of bank
borrowers around the world are unlisted I We examine the role of (the lack of) market information
in the ratings process I Market information disciplines rating agencies to provide more accurate
and timely ratings I The lack of market discipline gives rating agencies discretion to change their
methodologies and stringency I Regulators should reconsider the decision to allow banks to use
credit ratings for capital allocation I Rating agencies should make the ratings of listed and unlisted
firms more comparable
 LITERATURE SURVEY

• Fintech and credit scoring for the millennial generation

• 1. Digital footprint data has a significant correlation with the loan approval probability. Those with
a significant digital mobile presence are significantly more likely to get approved 2. Digital and
Social footprint variables have higher (relative to credit score) predictive power in predicting
defaults • as compared to the credit score, the digital footprint variables taken together are able to
explain 10 percentage points higher variation in defaults. AUC of credit score is 58.6% and of
social footprint variables alone is 69% 3. The discriminatory ability of the digital footprint
variables varies based on the loan purpose Customers who log in via Facebook are 20%, 26%, and
32% more likely to default when they take loans for making a purchase, meeting the EMI on an
existing loan, or repayment of an existing loan respectively but less likely to default when they
take a loan for medical needs.
 LITERATURE SURVEY

• Algorithmic traders and volatility information trading

• I Non-algorithmic traders are informed regarding future volatility while algorithmic traders are not.
I The predictive ability of options market volatility demand rarely lasts more than two days into the
future. I Neither propitiatory (who trade in their own account) nor agency (who execute trades on
someone else’s behalf) algorithmic traders have volatility related information. I Both scheduled
and unscheduled corporate announcements act as exogenous shocks, resulting in volatility spikes.
Traders behave similarly in periods leading up to both these type of corporate announcements
 Proposed system

The machine learning model will be chosen from different algorithms according to
the best accuracy
More advanced techniques like ARIMA (auto regressive integrated moving
average) will be used.
This performs better with time series data than other algorithms as it was developed
only to handle time series data.
 Modules

• Python -- 3.7
• SkLearn – 0.7
• Tensorflow -- 2.2
• Numpy
• Pandas
• Matplotlib
System Architecture
 Outcome of the project

The developed model will be able to predict the stock price will go down or up the
next day it will also predict how much the stock will cost the next day. This can also
forecast to n days ahead of today, the caveat in this is increasing days will be taking
more forecasted values into the model which may diminish the final model
accuracy.
 End user

With this project Fund managers, Finance managers, Stock brokers and individuals
will be able to assess the risks and take better decisions according to the output of
the model.
 References

1. M. Usmani, S. H. Adil, K. Raza and S. S. A. Ali, "Stock 1. H. L. Siew and M. J. Nordin, "Regression techniques for
market prediction using machine learning the prediction of stock price trend," 2012 International
techniques," 2016 3rd International Conference on Conference on Statistics in Science, Business and
Computer and Information Sciences (ICCOINS), Kuala Engineering (ICSSBE), Langkawi, 2012, pp. 1-5.
Lumpur, 2016, pp. 322-327. 2. K. V. Sujatha and S. M. Sundaram, "Stock index
2. K. Raza, "Prediction of Stock Market performance by prediction using regression and neural network models
using machine learning techniques," 2017 International under non normal conditions," INTERACT-2010,
Conference on Innovations in Electrical Engineering and Chennai, 2010, pp. 59-63.
Computational Technologies (ICIEECT), Karachi, 2017, 3. S. Liu, G. Liao and Y. Ding, "Stock transaction prediction
pp. 1-1. modeling and analysis based on LSTM," 2018 13th IEEE
3. H. Gunduz, Z. Cataltepe and Y. Yaslan, "Stock market Conference on Industrial Electronics and Applications
direction prediction using deep neural networks," 2017 (ICIEA), Wuhan, 2018, pp. 2787-2790.
25th Signal Processing and Communications
Applications Conference (SIU), Antalya, 2017, pp. 1-4.
4. M. Billah, S. Waheed and A. Hanifa, "Stock market
prediction using an improved training algorithm of
neural network," 2016 2nd International Conference on
Electrical, Computer & Telecommunication Engineering
(ICECTE), Rajshahi, 2016, pp. 1-4.
 References

8. T. Gao, Y. Chai and Y. Liu, "Applying long short term momory neural networks for predicting stock closing price," 2017 8th IEEE International
Conference on Software Engineering and Service Science (ICSESS),Beijing, 2017, pp. 575-578.
9. Desai. Mitesh Madhusudhan Stock Closing Price Prediction using Machine Learning SVM ModelInternational Journal For Research In Applied
Science & Engineering Technology, 2020
10. K. A. Althelaya, E. M. El-Alfy and S. Mohammed,Evaluation of bidirectional LSTM for short-and longterm stock market prediction," 2018 9th
International Conference on Information and Communication Systems (ICICS), Irbid, 2018, pp. 151-156.
11. Mispricing in single stock futures: Empirical examination of Indian markets R.L. Shankar, Ganesh Sankar, Kiran Kumar K.
12. How Good is the Transformation-Based Approach to Estimate Value at Risk? Simulation and Empirical Results G.P. Samanta.
13. How Good are Bank Loan Ratings in India for Listed and Unlisted Firms? Yadav Gopalan, Indiana University Radhakrishnan Gopalan, Washington
University Kevin Koharki, Purdue University
14. Fintech and credit scoring for the millennial generation Alok Shashwat, Indian School of Business Sumit Agarwal, National University of Singapore
Pulak Ghosh, IIM Bangalore Sudip Gupta, Fordham University.
15. Algorithmic traders and volatility information trading Anirban Banerjee, IIM Kozhikode Ashok Banerjee, IIM Calcutta.