Stock Price Prediction and Alert System

Abstract

In the financial markets, accurate prediction of stock prices is important for investors seeking to optimize returns. This paper presents a project focused on Stock Price Prediction Using LSTM and Alert System. The Long Short-Term Memory (LSTM) models, known for their ability to capture intricate patterns in sequential data, our project aims to forecast stock prices with enhanced accuracy and reliability. Additionally, we introduce an alert system. This supplementary feature enables users to set alerts based on specific stock price, enhancing timely responses to market changes. The methodology involves rigorous data analysis using Yahoo Finance data, preprocessing techniques, and the implementation of LSTM models for predictive modeling. Through comprehensive evaluation and experimentation, we demonstrate the efficacy of LSTM models in generating accurate stock price predictions.

Introduction

I. INTRODUCTION

The dynamics of the stock market, with its intricate interplay of economic indicators, investor sentiments, and global events, have made predicting stock prices an increasingly challenging yet essential endeavour. Investors seek tools and methodologies that can provide insights into market trends, aiding in informed decision-making to optimize returns on investments. In recent years, the advent of predictive analytics, fueled by the availability of vast datasets, has spurred interest in leveraging artificial neural networks (ANNs) to unlock patterns within this complex financial ecosystem.

Our focus on stock market prediction stems from the recognition that the profitability of investors hinges on their ability to anticipate market movements. Traditional methods have given way to more sophisticated algorithms capable of handling sequential data, and among them, recurrent neural networks (RNNs) have demonstrated notable potential. The application of RNNs, such as Gated Recurrent Units (GRUs), has been explored in recent works. However, inherent challenges, notably the high loss rates associated with GRUs, have prompted us to delve into alternative approaches to enhance the accuracy and performance of stock market predictions. In this pursuit, we turn our attention to Long Short-Term Memory (LSTM), a specialized form of RNN. The unique architecture of LSTMs addresses the persistent challenge of error propagation through time and layers, making them particularly suitable for modeling sequential data.

Additionally, we introduce a supplementary feature, the stock alert system, which signals the likelihood of stock price movements. While not the central focus, this side feature offers an additional layer of utility for investors keen on staying informed and responsive to market changes.

In the subsequent sections, we delve into the nuances of our proposed methodology, the experimental setup using Yahoo Finance data, and the comprehensive evaluation of results.

Through this research, we aim to shed light on the efficacy of LSTM in stock market prediction, providing valuable insights for investors navigating the ever-evolving financial markets.

II. LITERATURE SURVEY

From the literature survey, it was observed that the machine learning techniques are proving much more accurate than the other prediction techniques Accurately predicting future trends is essential for managing crises and making profits in unpredictable financial markets, such as stocks. Radu Iacomin[1] study show that advanced machine learning methods, especially SVM with PCA feature selection, are useful for forecasting nonlinear signals and maximizing profitability. Sumeet Sarode , Harsha G. Tolani , Prateek Kak, Lifna C [2] performed Stock Market Prediction. The literature emphasises how difficult it is to predict the stock market because of its volatile and dynamic environment. The study presents an integrated strategy that uses LSTM for price prediction and real-time news analysis to capture investor sentiments, incorporating insights from behavioural finance and providing thorough recommendations for future investment decisions. Rachna Sable Dr. Shivani Goel , Dr. Pradeep Chatterjee [4] did Empirical Study on Stock Market Prediction Using Machine Learning .This paper aims to study the stock   market prediction using multiple Traditional, Machine learning, and Deep learning algorithms.

Along with the algorithms, the survey has focused on various datasets used for stock market prediction, features of these datasets selected as input parameters and the evaluation metrics used for comparing the results of predictions. Warren Landis and Sangwhan Cha[5] studied High Performance Stock Market Prediction Methods. The literature highlights how timely and efficient transactions are essential for stock markets to generate maximum profits, which encourages investors to investigate predictive machine learning systems. The study acknowledges the difficulties in obtaining sufficient data and suggests an ensemble learning strategy that uses Long Short-Term Recurrent Neural Networks (LSTM) to improve the timeliness of stock predictions by experimenting with a variety of big data sources. Sneh Kalra and Jay Shankar Prasad [6] studied efficacy of News Sentiment for Stock Market Prediction Because stock markets are stochastic, the literature emphasises how difficult it is to predict stock market trends. The research, which makes use of a wealth of data from various sources, suggests a daily prediction model that incorporates historical data and news articles, applies sentiment analysis, and uses machine learning techniques to achieve accuracy ranging from 65.30 to 91.20.Gourav Bathla [3] predicted stock price using LSTM and SVR .The study compares the effectiveness of Support Vector Regression (SVR), a conventional method, and Long Short Term Memory (LSTM), a deep learning technique, in order to address the difficulties in forecasting non-linear and complex stock price movements. Using historical data from multiple stock indices, the analysis evaluates metrics such as Mean Absolute Error to determine how accurate the predictions are. The purpose of the study is to compare how well LSTM captures the complex patterns of stock prices compared to SVR.

III. METHODOLOGY

Stock market prediction poses a multifaceted challenge, requiring developers and researchers to address numerous factors. Leveraging machine learning methods facilitates the establishment of connections between historical and current data, empowering machines to learn and generate accurate predictions.

A. Data Analysis Stage

The foundation of our methodology lies in robust data analysis, recognizing the critical role it plays in the accuracy of predictions. Utilizing machine learning hinges on the seamless correlation of past and present data, enabling the system to discern patterns and trends.

For our study, the Yahoo Finance API serves as the primary tool for extracting both historical and real-time data. The dataset encompasses essential variables such as open, close, low, high, volume, and adjacent close. To ensure the dataset’s reliability, we implement the MinMaxScaler to scale the selected features within a specified range, mitigating the impact of minor data fluctuations that could otherwise lead to substantial variations in predictions. This meticulous data preparation stage establishes a solid foundation for subsequent machine learning processes.

B. Flowchart

Conclusion

The proposed system introduces a robust stock price prediction methodology leveraging LSTM models. Although it is impossible to predict a stock\'s exact price, a deep learning model can help by predicting the stock value based on past values. This can assist us in assessing the direction or the state of the market. Validated with Yahoo Finance data and includes a practical stock alert system. The results highlight LSTM\'s superior accuracy, offering a promising tool for precise stock predictions in dynamic financial markets.

References

[1] Radu Iacomin, “Stock Market Prediction”, 2015 19th International Conference on System Theory, Control and Computing (ICSTCC), October 14-16, Cheile Gradistei, Romania. [2] Sumeet Sarode, Harsha G. Tolani, Prateek Kak, Lifna C , “STOCK PRICE PREDICTION USING MACHINE LEARNING TECHNIQUES”, International Conference on Intelligent Sustainable Systems (ICISS 2019) IEEE Xplore Part Number: CFP19M19-ART; ISBN: 978-1-5386-7799-5. [3] Gourav Bathla, “Stock Price prediction using LSTM and SVR”, 2020 Sixth International Conference on Parallel, Distributed and grid computing(PDGC). [4] Rachna Sable , Dr. Shivani Goel , Dr. Pradeep , “Empirical Study on Stock Market Prediction Using Machine Learning”. [5] Warren Landis, Sangwhan Cha, “Towards High Performance Stock Market Prediction Methods”, 2020 IEEE Cloud Summit. [6] Sneh Kalra , Jay Shankar Prasad, “Efficacy of News Sentiment for Stock Market Prediction”,2019 International Conference on Machine Learning, Big Data, Cloud and Parallel Computing (Com-IT-Con), India, 14th -16th Feb 2019. [7] Shi yan, “Understanding LSTM and its Diagrams”,Mar14 2016.

Copyright

Copyright © 2024 Harshati Farde, Neha Deore, Dhanashree Manyar, Subodh Magar, Mr. N. R. Khaire. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.