Pradip Gareja et al; International Journal of Advance Research and Development

(Volume 3, Issue 10)

Available online at: www.ijarnd.com
Stock market prediction using machine learning
Gareja Pradip1, Chitrak Bari2, J. Shiva Nandhini3
1,2
Student, SRM Institute of Science and Technology, Chennai, Tamil Nadu
3
Assistant Professor, SRM Institute of Science and Technology, Chennai, Tamil Nadu

ABSTRACT The three different approaches for solving the problem:

Fundamental analysis, Technical Analysis and The application
The basic tool aimed at increasing the rate of investor's interest of Machine Learning. We found evidence in support of the weak
in stock markets is by developing a vibrant application for form of the Efficient Market Hypothesis. We can use
analyzing and predicting stock market prices. In this report we Fundamental Analysis and Machine Learning to guide an
explain, the development and implementation of a stock market investor’s decisions.
price prediction application using a machine learning
algorithm. In this report, we try to analyze existing and new 2. FUNDAMENTAL ANALYSIS
methods of stock market prediction. We take three different
approaches to solving the problem: Fundamental analysis,
Technical Analysis and The application of Machine Learning.
We found evidence in support of the weak form of the Efficient
Market Hypothesis. We can use Fundamental Analysis and
Machine Learning to guide an investor’s decisions. We
demonstrate a common flaw in Technical Analysis
methodology to show that it produces limited useful
information. Based on our findings, algorithmic trading
programs are developed and simulated using Quant. During
the past few decades, various machine learning techniques
have been applied to study the highly theoretical and
speculative nature of the stock market by capturing and using
repetitive patterns. Different companies use different types of Fig. 1: The Proposed fundamental analysis-based approach
analysis tools for forecasting and the main aim is the accuracy, to stock market forecasting
with which they predict which set of stocks would yield the
maximum amount of profit. The fundamental analysis of stocks is the essential of investing–
and the foundation of most of the strategies covered in this
Keywords— Stock market, Application development, Machine tutorial. It involves using quantitative and qualitative factors to
learning algorithm, Python, Reinforcement learning
answer questions such as:
1. INTRODUCTION  Are the company’s revenues really growing?
It is a place, where shares of public listed companies are traded  Is the company actually making any profit?
for buying and selling purposes. It is the place, where the stock
 Can the company successfully beat its competitors in the
buyers and sellers meet. The stock Market prediction has always
future?
been appealing for the researchers. There have been numerous
 Can the company repay its debts if such a condition arose in
scientific attempts to predict the stock market effectively but no
near future?
method has been discovered accurately to predict stock price
movement. Stock market prediction is trying to determine the  And ultimately: Will this company’s stock be a good
investment?
future value of a company stock or other financial instrument
traded on an exchange. To successfully predict stock's future  This aim is to evaluate the value of the underlying company.
price could yield a significant profit. The efficient-market It takes into account, the genuine value of the share keeping
hypothesis suggests us that stock prices reflect all currently in mind the economic conditions and the industry along with
available information and any price changes that are not based the company’s financial condition and management
on the newly revealed information thus are inherently performance. A fundamental analyst would definitely have
unpredictable. To determine future forecasts, fundamentalists a detailed glance at the balance sheet, the profit and loss
use numeric information such as earnings, ratios, and statement, financial ratios and other data that could be used
management effectiveness. In the Technical analysis, market to predict the future of a company. In other words, the
timing is believed to be the key. To identify trends in price, fundamental stock market analysis is about using real data
technicians utilize charts and modeling techniques. Then they to evaluate a stock's value. So they go for a detailed study
rely on historical data in order to predict future outcomes. on revenues, earnings, future growth, return on equity, profit
© 2018, www.IJARND.com All Rights Reserved Page | 29
 Pradip Gareja et al; International Journal of Advance Research and Development
margins, and other data to determine a company's 3. TECHNICAL ANALYSIS
underlying value and potential for future growth.  High Liquidity- Liquidity is an essential volume. Heavily-
 The basic belief is that, as the company will grow, so will traded stocks allow the investors to trade quickly and easily,
the value of the share. So this will, in turn, benefit the without dramatically changing the price of the stocks. Thinly-
investor in the long run. traded stocks are generally difficult to trade, because there
aren't many buyers or sellers at any given time, so buyers and
2.1 Fundamental analysis-based stock market forecasting sellers may have to change their desired price considerably in
techniques order to make a trade. In addition, low liquidity stocks are
This section describes the techniques used in fundamental highly low priced (sometimes less than a penny per share),
analysis-based stock market forecasting, including weight which means that their prices can be more easily manipulated
calculation of financial indicators, individual stock evaluation by individual investors. These outside forces that acting on
and selection, feature selection of financial news, stock trading thinly-traded stocks make them quite unsuitable for technical
signal determination based on financial news, and stock price analysis.
trend forecasting, all of which are further discussed below.  No Artificial Price Changes- Splits, dividends, and
distributions are the most common “culprits” for the artificial
A. Weight calculation of financial indicators price changes. Though there is no noticeable difference in the
The firm financial condition is one of the key factors involved in value of the investment, artificial price changes can
investment decisions. Evaluating firm financial condition dramatically affect the price chart and make technical
requires first capturing the financial indicators of individual analysis difficult to apply. This kind of price influence from
stocks from the database of the Taiwan Economic Journal (TEJ). the outside sources can be easily addressed by adjusting the
These financial indicators are then normalized. Finally, weights historical data prior to the price change.
of financial indicators are calculated to pave the way for  No Extreme News- Technical analysis cannot predict the
individual stock evaluation and selection. extreme events, including business events such as a
company's CEO dying unexpectedly, and political events
B. Individual stock evaluation and selection such as a terrorist act. When the forces of “extreme news” are
Based on the weights of normalized financial indicators, this highly influencing the price, technicians have to wait very
section evaluates firm financial condition to select optimal patiently until the chart settles down and starts to reflect the
stocks and predict their future price trends. “new normal” that results from such news.

C. Feature selection of financial news The basic of technical analysis:

Financial news can strongly affect the prices of individual  Price Discounts Everything
stocks. To evaluate the influence of financial news on individual  Price Movements Are Not Totally Random
stock prices, this section identifies the features from the financial
news that affect the individual stock price. A. Technical analysis-based stock market forecasting
Process This section designs the process of technical analysis-
D. Stock trading signal determination based on financial news based stock market forecasting to help investors effectively and
In data clustering techniques, the Naïve Bayesian classifier [7] accurately predict the trends of individual stock through the
exhibits the best learning efficiency and accuracy. This study technical analysis for use in stock market trading, as shown in
thus applies the Naïve Bayesian classifier to classify financial Fig. 1. The technical analysis-based stock market forecasting
news to clarify its effects on individual stock price. The process primarily includes the following three phases: trend
financial news is considered the input variable of the Naïve based stock classification, adaptive stock market indicator
Bayesian classifier, while the effect of financial news on stock selection and stock market trading signal forecasting.
trading signal is considered the output variable of the Naïve
Bayesian classifier. The process comprises two steps, namely
establishment of financial news dataset and classification of
financial news.

Fig. 2: Shows you how we can predict stock price using this
method.in this fig, we can see the stock price is many times goes
up and down so we can easily predict stock price using this Fig. 3: Technical analysis-based stock market forecasting
method. process

© 2018, www.IJARND.com All Rights Reserved Page | 30

 Pradip Gareja et al; International Journal of Advance Research and Development
The MLP network has three layers: the input layer, an output
layer, and a hidden layer. Neuron takes the values of inputs
parameters, sums them up according to the assigned weights, and
then adds a bias. By applying the transfer function, the value of
the outputs could be determined. The number of neurons in the
input layer corresponded to the number of input parameters.

Artificial neural networks (ANNs) are an information processing

system that was first inspired by generalizations of mathematical
of the human neuron (Figure 5).

Fig. 5: Architecture of a feed-forward MLP

Each neuron receives some signals from other neurons or

outside. Above figure has three layers of neurons, where one
input layer is present. Every neuron employs activation function
that fires when the total input is more than a given threshold.

3.1.2 Multiple linear regression: Regression is used for

Fig. 4: Algorithm for establishing associated interval predicting an outcome based on a given input. The simplest
sequence regression technique called linear regression and the advanced
one called multiple regression. It depends on the variable. If a
3.1 Machine Learning approaches for stock market single descriptive variable is being used, it is known as simple
prediction. linear regression and if more than one descriptive variable is used
Artificial Intelligence (AI) and Machine Learning (ML) are quite when the technique is multiple regression. The multiple linear
revolutionizing nearly all areas of our lives. Did you know, that regression predicts the future value of the variable (Yˆ) with
the use of Machine Learning for trading is getting more and more respect to other variables (Xi) using equation 1.
important? You might be surprised to know that Machine
Learning hedge funds already significantly outperform 𝑌̂ = 𝛽0 + 𝛽1 𝑋1 + 𝛽2 𝑋2 + ⋯ + 𝛽𝑛 𝑋𝑛 (1)
generalized hedge funds, as well as traditional quant funds,
according to a report by Value Walk. ML and AI systems are Where 𝛽0 , 𝛽1 , 𝛽2 … 𝛽𝑛 are co-efficient that can be calculated usinf
incredibly helpful tools for humans navigating the decision- equation 2.
making process involved with investments and risk assessment. 𝛽 = (𝑋 𝑇 𝑋)−1 𝑋 𝑇 𝑌 (2)
The impact of human emotions on trading decisions is often the This multiple linear regression will be hybrid by moving average
greatest hindrance for outperformance. Algorithms and to avoid data outlier [5]. Moving is an arithmetic moving average
computers make better decisions and execute trades faster than calculated by adding the closing price of the security for a
any human can, and do so free from the influence of emotions. number of time periods and then dividing this total by the
There are many different types of algorithmic trading. A few number of time periods. Moving average is applied based
examples are as follows: equation 3.
 Trade execution algorithms, which break up trades into ∑𝑛 𝑋
smaller in order to minimize the impact on the stock price. ̅𝑋̅̅𝑛̅ = 𝑖=1 𝑖 (3)
𝑛
An example of this algorithm is a Volume Weighted
Average Price (VWAP) strategy. 3.1.3 Bayesian algorithm: After analyzing the entire learning
 Trade execution algorithms, which break up trades into algorithm the Naive Bayes classifier is being selected for the
smaller in order to minimize the impact on the stock price. following reasons.
An example of this algorithm is a Volume Weighted  There is no need to search through all the version space for
Average Price (VWAP) strategy. the optimal hypothesis in this algorithm.
 Stealth/gaming algorithms that are used for detecting and  Space for the optimal hypothesis in this algorithm ).- Prior
taking advantage of price movements caused by large trades knowledge can be effectively used in this method. Often in
and/or other algorithm strategies. the share market, some trend can be found to be prominent
and good to use as a valid a priori probability which can help
3.1.1 Artificial neural network: A neural network is a type of the learner to show better performance.
bio-inspired system with several single processing elements,  Simple to implement and efficiently usable for a large
called neurons. The neurons are connected to each other by a number of attributes. ). Further training can be done for new
joint mechanism which is consisted of a set of assigned weights. training instances by updating the probability, received from
MLP is a very common approach in regression-type problems. the previous training.
© 2018, www.IJARND.com All Rights Reserved Page | 31
 Pradip Gareja et al; International Journal of Advance Research and Development
The Bayes Naive classifier selects the most likely classification instruments that might help in arriving at a better outcome for
Vnh given the attribute values aj, a2, . . ..an These results in, decision-making.
arg 𝑚𝑎𝑥 𝑎𝑖 • Convenience galore: One of the main factors to choose the
𝑉𝑛𝑏 = 𝑉𝑓 𝑃(𝑉𝑗 )𝜋𝑃( 𝑉𝑗 ) (4) perfect software who has the ability to have trend analysis
within a matter of minutes in your hand.
𝑎
We generally estimate 𝑃( 𝑖 ) using m-estimates:
𝑉𝑗
5. CONCLUSION
𝑎𝑖 𝑛𝑐 + 𝑚𝑝 (5) The aim of our research study is to help the stockbrokers and
𝑃( ) =
𝑉𝑗 𝑐+𝑚 investors in investing money in the stock market. The prediction
plays a very important role in stock market business which is
Where, n= Number of training examples where v = Vi nc = very complicated and challenging process due to the dynamic
number of examples where v = Vi and a = aj p = a priori estimate nature of the stock market.
for P (aj I Vi ) m = the equivalent sample size
6. REFERENCES
4. PROPOSED SYSTEM [1] Krishna Kumar Singh, Dr, Priti Dimri, Madhu Rawat:”
Dept. of Computer Science, GBPEC, Ghurdauri, “Green
Database Model for Stock Market: a Case Study of Indian
Stock Market”.
[2] Ashish Sharma, Dinesh Bhuriya, Upendra Singh: Women’s
Polytechnic, Indore ” Survey of Stock Market Prediction
Using Machine Learning Approach”, International
Conference on Electronics, Communication and Aerospace
Technology ICECA 2017.
[3] Carol Hargreaves, Yi Hao: ” Does the use of Technical &
Fundamental Analysis improve Stock Choice? : A Data
Mining Approach applied to the Australian Stock Market”
[4] Abidal Izzah, Yuita Arum Sari, Ratna Widyastuti, Toga
Aldila Cinderatama: “Mobile App for Stock Prediction
Using Improved Multiple Linear Regression”,” 2017
Fig. 6: here we design an architectural diagram of the stock International Conference on Sustainable Information
market prediction system Engineering and Technology (SIET)”.
[5] Yi Zuo, Masaaki Harada, Takao Mizuno, Eisuke Kita:
4.1 Advantages of proposed system ”Application of Byyeian for Nikkei stock Return
• Biases, gone forever: When it comes to a human element Prediction”, ”2011 Conference on Technology and
within any analysis, it’s difficult to avoid biases. Investing in Applications of Artificial Intelligence”.
the stock market can mean that one is investing in emotions [6] A Sheta, “Software Effort Estimation and Stock Market
that influence the decision to withdraw investments or not, so Prediction Using Takagi-Sugeno Fuzzy Models”, In
we tried hard so that biases don’t occur. Proceedings of The IEEE International Conference on
• Cross-checking results become important: The quotes on Fuzzy Systems, pp.171- 178, Vancouver, BC, 2006.
the stock market are based on market conditions and how the [7] M.H. FazelZarandi, B. Rezaee, I.B. Turksen, and E. Neshat,
market works, but they can be predicted with the help of “A type-2 fuzzy rule-based expert system model for stock
certain software that gives predictive results of the outcomes price analysis”, Expert Systems with Applications, Vol.36,
of these market trends. Once the information with the given No.1, pp. 139-154, January 2009.
factors that impact the market is fed in, the requisite [8] Robert K. Lai, Chin-Yuan Fan, Wei-Hsiu Huang, and Pei-
fluctuations are also factored in, giving a result that is near- Chann Chang, “Evolving and clustering fuzzy decision tree
accurate to the actual market conditions. for financial Time series data forecasting”, An International
• Scheduling tasks: Our software helps you (investor) conduct Journal of Expert Systems with Applications, Vol.36, No.2,
a variety of tasks that can be easily scheduled without fail. pp. 3761-3773, March 2009.
With these schedules we are able to initiate tasks without any [9] Shyi-Ming Chen and Yu-Chuan Chang, “Multi-Variable
delays, thus conducting our analysis on time for your better Fuzzy Forecasting Based On Fuzzy Clustering and Fuzzy
outcome. Rule Interpolation Techniques”, Information Sciences,
• Consistency assured: Human indulgence often restricts the Vol.180, No.24, pp. 4772-4783, 2010.
outcome of the results. Often biases or human delays reduce [10] S Abdulsalam Sulaiman Olaniyi, Adewole, Kayode S.,
the consistency of the results as expected before. With our Jimoh, R. G,” Stock Trend Prediction Using regression
stock market software by your side, one is capable of simply analysis – A Data Mining Approach”, ARPN Journal of
running the results based on the latest data that is fed-in or Systems and Software Volume 1 No. 4, July 2011, Brisbane,
directly extracted from sources. Australia.
• Don’t hesitate to conduct a thorough analysis: It is quite
possible while conducting an analysis, we ignore the essential

© 2018, www.IJARND.com All Rights Reserved Page | 32