International Journal of All Research Education and Scientific Methods (IJARESM),

ISSN: 2455-6211, Volume 13, Issue 5, May-2025, Available online at: www.ijaresm.com

Stock Price Prediction Using Machine Learning

Badal Bhushan1, Kunal Kumar2, Manish Kumar Srivastava3,
Nitish Narayan Singh4, Priyanshu Sharma5
1
Assistant Professor, Department of Computer Science and Engineering, IIMT College of Engineering, Greater Noida,
India
2,3,4,5
B. Tech (CSE) Research Scholar, Department of Computer Science and Engineering, IIMT College of
Engineering, Greater Noida, India

-------------------------------------------------------------****************-----------------------------------------------------------

ABSTRACT

Stock market prediction has been a subject of extensive research due to its financial significance and complexity.
The unpredictable nature of stock price movements, influenced by various macroeconomic, political, and
psychological factors, presents a challenge for investors and analysts. Traditional forecasting techniques such as
fundamental and technical analysis have been widely used, but recent advancements in machine learning have
significantly improved prediction accuracy.This study explores the application of machine learning techniques,
including regression models, decision trees, random forests, and deep learning architectures like Long Short-
Term Memory (LSTM) networks, to forecast stock prices based on historical data and market sentiment
analysis. We employ feature engineering techniques such as moving averages, volatility indices, and sentiment
scores derived from financial news and social media. Additionally, statistical tests like the Augmented Dickey-
Fuller (ADF) test are used to analyze market trends and stationarity.

Our methodology includes data collection from sources such as Yahoo Finance and Kaggle, preprocessing to handle
missing values, and hyperparameter tuning using GridSearchCV to optimize model performance. The models are
evaluated using performance metrics such as Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), and R-
squared (R²) score. Results indicate that deep learning models outperform traditional machine learning techniques due
to their ability to capture sequential dependenciesdecision-making in stock price movements.Despite the promising
results, challenges such as overfitting, data scarcity, and market volatility persist. Future research should explore hybrid
models integrating reinforcement learning and real-time data processing to enhance predictive accuracy. This study
provides valuable insights for investors and tradersseeking data-driven decision-making tools.

INTRODUCTION

Stock market prediction has always been a critical area of research due to its impact on financial markets and
investment strategies. Investors, analysts, and financial institutions seek accurate models to forecast stock prices and
trends, which can lead to profitable decision-making and risk mitigation. However, the stock market is inherently
volatile and influenced by multiple factors such as economic indicators, geopolitical events, interest rates, corporate
performance, and market sentiment. This unpredictability makes stock price forecasting an immensely challenging
task.Traditional stock market forecasting approaches are primarily categorized into fundamental analysis and technical
analysis. Fundamental analysis focuses on evaluating a company's financial health, including revenue, earnings, assets,
liabilities, and economic conditions, to determine its intrinsic value. Technical analysis, on the other hand, relies on
historical stock price patterns, volume trends, and various technical indicators such as moving averages, Bollinger
Bands, and Relative Strength Index (RSI). While these methods have been widely used, they often fail to account for
the dynamic and non-linear nature of financial markets.

Another emerging area in stock price prediction is sentiment analysis, where Natural Language Processing (NLP)
techniques analyze news articles, financial reports, and social media discussions to determine market sentiment. Studies
have shown that market sentiment can significantly influence stock prices, as positive or negative news can lead to
corresponding changes in investor behavior. Combining sentiment analysis with machine learning-based forecasting
models has further improved prediction accuracy.Despite these advancements, several challenges persist in stock price
prediction. Overfitting is a common issue in deep learning models, where the model learns patterns too specific to
training data, leading to poor generalization on unseen data. Additionally, market anomalies and black swan events,
such as financial crises and political instability, can cause sudden and unpredictable fluctuations in stock prices. Data
availability and quality also play a crucial role, as missing or inconsistent data can adversely affect model performance.
This research aims to develop a robust stock price prediction model by integrating multiple machine learning
techniques and evaluating their effectiveness. We utilize historical stock data, technical indicators, and sentiment
analysis to enhance forecasting accuracy. The study also assesses the strengths and limitations of different models and

Page | 947
 International Journal of All Research Education and Scientific Methods (IJARESM),
ISSN: 2455-6211, Volume 13, Issue 5, May-2025, Available online at: www.ijaresm.com

provides insights into their practical applications for investors and traders.In the following sections, we review existing
literature on stock price prediction, outline the methodology used in our research, analyze experimental results, discuss
challenges and limitations, and propose future directions for improving predictive models. This research contributes to
the growing field of AI-driven financial forecasting and offers a structured approach to understanding stock market
dynamics using machine learning.

LITERATURE REVIEW

In recent years, researchers have explored various machine learning and deep learning techniques for stock price
prediction, leveraging advancements in artificial intelligence, big data analytics, and sentiment analysis.This section
reviews ten recent studies published after 2020 that have contributed significantly to this domain.

Chen et al. (2021) proposed a hybrid model combining Long Short-Term Memory (LSTM) networks with Gated
Recurrent Units (GRU) to improve stock price prediction accuracy. The study demonstrated that deep learning models
could capture sequential patterns in financial time-series data more effectively than traditional models.

Weng et al. (2021) examined the effectiveness of transformer-based models for stock price prediction. The research
highlighted that attention mechanisms significantly improved forecasting by capturing long-range dependencies in
stock movements. Their findings showed that transformer models outperformed LSTMs in certain volatile market
conditions.

Xu et al. (2022) investigated the integration of sentiment analysis with machine learning-based stock price prediction.
The study used natural language processing (NLP) techniques to analyze financial news and social media data,
revealing that sentiment scores had a strong correlation with stock price fluctuations.

Ghosh & Roy (2022) developed an ensemble learning approach using Random Forest, XGBoost, and LightGBM to
predict stock market trends. The study demonstrated that combining multiple models improved robustness and reduced
overfitting, leading to higher prediction accuracy.

Kim et al. (2023) explored the impact of macroeconomic indicators such as inflation, interest rates, and GDP growth
on stock prices. They developed a multi-variable regression model that incorporated economic data, showing improved
predictive performance compared to models based solely on historical stock prices.

Zhang & Li (2023) proposed a hybrid model that integrated ARIMA (AutoRegressive Integrated Moving Average)
with deep learning networks. The study found that while ARIMA captured short-term trends effectively, deep learning
models provided better long-term predictions by learning complex non-linear patterns.

Singh et al. (2023) investigated the use of reinforcement learning for stock trading and price prediction. Their study
implemented Deep Q-Networks (DQN) to optimize buy and sell decisions, showing that reinforcement learning models
could adapt to changing market conditions better than static machine learning models.

Wang et al. (2024) introduced a novel convolutional neural network (CNN)-based approach for stock price prediction.
By transforming time-series data into image-like representations, CNNs were able to extract important features and
achieve competitive accuracy compared to traditional recurrent networks.

Patel & Sharma (2024)analyzed the effectiveness of Federated Learning (FL) in financial predictions. Their study
highlighted that FL allows multiple financial institutions to collaboratively train models without sharing sensitive data,
improving privacy while maintaining high predictive performance.

Chen & Huang (2024) presented an advanced deep reinforcement learning model that incorporated real-time data
streaming for high-frequency trading applications. Their results indicated that models trained with real-time updates
outperformed static models in fast-changing market environments.

METHODOLOGY

The methodology for stock price prediction involves several stages, including data collection, feature engineering,
model development, evaluation, and optimization. In this research, the objective is to combine both traditional time-
series forecasting methods and modern machine learning (ML) techniques to improve stock price prediction accuracy.
The following steps outline the methodology used in this project:

Page | 948
 International Journal of All Research Education and Scientific Methods (IJARESM),
ISSN: 2455-6211, Volume 13, Issue 5, May-2025, Available online at: www.ijaresm.com

Data Collection and Preprocessing

The first step in the methodology is to gather historical stock price data. This data typically includes the opening,
closing, high, and low prices, as well as the trading volume for a given period (e.g., daily, weekly, monthly). Additional
financial indicators such as market sentiment, interest rates, and macroeconomic factors may also be incorporated. Data
can be sourced from stock exchanges or financial APIs like Yahoo Finance, Alpha Vantage, or Quandl.
Once the data is collected, preprocessing steps are applied to handle missing values, remove outliers, and normalize the
data to ensure consistent scaling. Data normalization helps improve the model's ability to process inputs and prevents
features with larger numerical ranges from dominating the learning process. Common normalization methods include
Min-Max scaling or Z-score normalization.

Feature Engineering
Feature engineering involves transforming raw data into meaningful variables that enhance the predictive power of the
model. In the case of stock price prediction, technical indicators are often used to capture trends, volatility, and
momentum. Some commonly used technical indicators include:

 Moving Averages (MA): Averages of stock prices over a specific period, such as Simple Moving Averages
(SMA) and Exponential Moving Averages (EMA).
 Relative Strength Index (RSI): A momentum oscillator that measures the speed and change of price movements.
 Bollinger Bands: A volatility indicator that uses a moving average and standard deviations to create upper and
lower bands around the price.
 MACD (Moving Average Convergence Divergence): An indicator that helps identify price trends and reversals.

These technical indicators, along with other financial data such as volume, are used as input features for the machine
learning model.

Model Development
The next step is selecting the machine learning models for stock price prediction. Given the nature of stock price data
(time-series and sequential), deep learning models like Long Short-Term Memory (LSTM) networks, which are
capable of capturing long-term dependencies, are widely used. In this project, we employ a hybrid model that combines
LSTM with traditional models such as ARIMA (Auto-Regressive Integrated Moving Average) for time-series
forecasting.

 LSTM: LSTM is a type of recurrent neural network (RNN) designed to handle sequential data and temporal
dependencies. LSTM is particularly useful for predicting stock prices, as it can learn from past price movements
and predict future trends while avoiding the vanishing gradient problem typically seen in traditional RNNs.
 ARIMA: ARIMA is a classical time-series forecasting model used to predict future values based on past data. It
models the relationship between an observation and a series of past observations by considering autoregressive
(AR), differencing (I), and moving average (MA) components.

A hybrid model is proposed, where ARIMA models the short-term trends and LSTM captures long-term dependencies
and complex patterns in the data.

Model Training and Optimization

The models are trained using historical stock data. The training dataset is divided into training and validation sets,
typically using an 80-20 split. For LSTM, the training process involves feeding sequences of stock prices (along with
technical indicators) into the network, allowing it to learn the patterns over time. Hyperparameter tuning is performed
to optimize the model's performance, including tuning the number of layers, units per layer, learning rate, and batch
size.

For ARIMA, the parameters (p, d, q) are tuned using techniques such as grid search or cross-validation to find the
optimal configuration.

Evaluation
After training, the model's performance is evaluated using a test dataset. Common evaluation metrics for stock price
prediction include:

 Mean Squared Error (MSE): Measures the average squared difference between predicted and actual values,
indicating how well the model performs.
 Root Mean Squared Error (RMSE): The square root of MSE, offering a more interpretable measure of
prediction accuracy.

Page | 949
 International Journal of All Research Education and Scientific Methods (IJARESM),
ISSN: 2455-6211, Volume 13, Issue 5, May-2025, Available online at: www.ijaresm.com

 Mean Absolute Percentage Error (MAPE): Provides the percentage error between predicted and actual
values, allowing for better insight into the relative performance.
 R-Squared (R²): Measures the proportion of the variance in the dependent variable that is predictable from the
independent variables.

Model Deployment and Real-time Prediction

Once the model is trained and evaluated, it is deployed for real-time prediction. This involves integrating the model
with a financial data source and using it to predict stock prices for the next day, week, or month. The system
continuously updates with new stock data to refine predictions. The model is tested with real-time data to ensure its
practical usability.

DATA ANALYSIS & RESULTS

Data Analysis
Before implementing predictive models, a thorough Exploratory Data Analysis (EDA) was
conducted to understand stock price trends, correlations, and key indicators influencing market
movements.

Dataset Overview
The dataset was sourced from Yahoo Finance and Kaggle, covering stock prices of multiple companies over a period of
5 to 10 years. It included the following features:

 Open Price: The first trading price of the stock for the day.
 Close Price: The final price at which the stock traded.
 High & Low Prices: The highest and lowest prices of the stock within a given timeframe.
 Volume: The number of shares traded during the day.
 Technical Indicators: Moving Averages, Relative Strength Index (RSI), Bollinger Bands.
 Sentiment Scores: Derived from financial news and social media analysis.

Data Preprocessing

 Handling Missing Values: Missing values were filled using interpolation and forward-fill techniques.
 Feature Scaling:MinMax Scaling was applied to normalize stock prices and prevent large-scale variations from
influencing the models.
 Stationarity Test: The Augmented Dickey-Fuller (ADF) test was used to check if the stock price data was
stationary. If non-stationary, differencing was applied to remove trends.
 Feature Engineering: Lag variables, moving averages, and volatility measures were generated to improve model
predictions.

Correlation Analysis
A correlation heatmap was generated to examine relationships between stock price movement and other variables. The
study found that closing price showed strong correlations with moving averages, Bollinger Bands, and sentiment
scores, confirming that these indicators could be valuable inputs for predictive models.

Model Implementation & Results

The dataset was split into 80% training and 20% testing, and multiple machine learning models were trained and
evaluated. The performance was measured using Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), and
R-squared (R²) score.x

Model Performance Comparison

Model MAE RMSE R² Score

Linear Regression 5.21 7.12 0.75
Decision Tree 4.89 6.45 0.78
Random Forest 3.67 5.12 0.85
Support Vector Machine (SVM) 4.12 5.98 0.82
LSTM Neural Network 2.34 3.89 0.92
GRU Neural Network 2.56 4.02 0.90

Page | 950
 International Journal of All Research Education and Scientific Methods (IJARESM),
ISSN: 2455-6211, Volume 13, Issue 5, May-2025, Available online at: www.ijaresm.com

Key Findings

 LSTM performed the best, with the lowest RMSE (3.89) and highest R² score (0.92), making it the most suitable
model for capturing stock price patterns.
 Random Forest and SVM also performed well, indicating that ensemble learning methods can provide good
approximations for stock price prediction.
 Linear Regression had the lowest accuracy, highlighting the limitations of simple regression models in capturing
non-linear stock price movements.

Sentiment Analysis Impact

By integrating sentiment analysis from financial news and social media, the accuracy of predictions improved by 5-7%.
Stocks associated with positive sentiment scores showed a higher probability of upward movement, whereas negative
sentiment correlated with price drops.

Discussion & Interpretation

The results confirm that deep learning techniques (LSTM, GRU) outperform traditional machine learning models in
stock price prediction due to their ability to recognize sequential patterns. Additionally, integrating technical indicators
and sentiment analysis enhances model accuracy by incorporating external market influences.
Despite these successes, challenges remain:

 Market fluctuations due to unexpected events (economic downturns, global crises).

 Deep learning models require significant computational resources and careful hyperparameter tuning.
 The availability and quality of real-time financial data affect prediction accuracy.

Performance Evaluation
To assess the effectiveness of different stock price prediction models, we used a combination of error metrics and
statistical measures. This section details the evaluation criteria, comparative analysis of models, and key findings.

Evaluation Metrics
The following performance metrics were used to evaluate the predictive accuracy of machine learning models:

1. Mean Absolute Error (MAE)

o Measures the average absolute difference between actual and predicted prices.
o Lower values indicate better model performance.
o
𝑛
1
𝑀𝐴𝐸 = 𝑦𝑖 − 𝑦𝑖
𝑛
𝑖=1

2. Root Mean Squared Error (RMSE)

o Penalizes larger errors more than MAE, providing a better estimate of model accuracy.
𝑛
1 2
𝑅𝑀𝑆𝐸 = 𝑦𝑖 − 𝑦𝑖
𝑛
𝑖=1

3. R-squared Score (R²)

o Represents the proportion of variance explained by the model.
o A score close to 1 indicates a highly accurate model.
𝑦𝑖 − 𝑦𝑖 2
𝑅2 = 1 −
𝑦𝑖 − 𝑦 2

4. Mean Squared Logarithmic Error (MSLE)

o Useful for financial forecasting where relative errors matter more than absolute differences.
𝑛
1
𝑀𝑆𝐿𝐸 = log 1 + 𝑦𝑖 − log 1 + 𝑦𝑖 2
𝑛
𝑖=1

5. Precision and Recall (for Sentiment Analysis-based Predictions)

o Precision: Measures how many of the predicted stock price increases were correct.

Page | 951
 International Journal of All Research Education and Scientific Methods (IJARESM),
ISSN: 2455-6211, Volume 13, Issue 5, May-2025, Available online at: www.ijaresm.com

o Recall: Measures how many actual price increases were correctly predicted.

Model Performance Comparison

The dataset was split into 80% training and 20% testing, and models were trained on historical stock data. Below is a
comparison of various models based on the evaluation metrics:

Model MAE RMSE R² Score MSLE

Linear Regression 5.21 7.12 0.75 0.052
Decision Tree 4.89 6.45 0.78 0.045
Random Forest 3.67 5.12 0.85 0.038
Support Vector Machine (SVM) 4.12 5.98 0.82 0.042

LSTM Neural Network 2.34 3.89 0.92 0.021

GRU Neural Network 2.56 4.02 0.90 0.024

Key Observations

 LSTM had the lowest MAE (2.34) and RMSE (3.89), making it the most accurate model.
 Random Forest and GRU also performed well, demonstrating the effectiveness of ensemble learning and
recurrent neural networks.
 Linear Regression performed the worst, highlighting its limitations in capturing non-linear stock price trends.
 Sentiment-based models improved accuracy by 5-7%, proving that external market sentiment influences stock
price movement.

Interpretation of Results

1. Deep learning models (LSTM, GRU) outperformed traditional machine learning models because they effectively
captured sequential dependencies in stock data.
2. Random Forest provided a good balance between accuracy and computational efficiency, making it a viable
alternative for quick predictions.
3. Hybrid approaches (Sentiment Analysis + Stock Price Prediction) improved model accuracy, highlighting the
importance of external factors.

Limitations & Challenges

Despite achieving high accuracy, the models faced certain challenges:

 Market Volatility: Unpredictable events (e.g., economic crises, political instability) can affect stock prices
unpredictably.
 Computational Complexity: Deep learning models require high computational power and extensive
hyperparameter tuning.
 Data Quality: Accuracy depends on high-quality, real-time financial data.

DISCUSSION

Key Findings
The results of this study demonstrate that machine learning models, particularly deep learning approaches like Long
Short-Term Memory (LSTM) networks, outperform traditional regression models in predicting stock prices. The
LSTM model effectively captures sequential dependencies and patterns in stock price movements, leading to lower
prediction errors compared to classical models such as Linear Regression, Decision Trees, and Random Forest.

Additionally, the study highlights the importance of feature engineering in improving prediction accuracy.
Incorporating technical indicators (e.g., moving averages, Bollinger Bands) and sentiment analysis of market news has
significantly enhanced model performance. The influence of macroeconomic indicators such as interest rates and
inflationwas also observed, suggesting that external economic factors play a crucial role in stock price movements.

Page | 952
 International Journal of All Research Education and Scientific Methods (IJARESM),
ISSN: 2455-6211, Volume 13, Issue 5, May-2025, Available online at: www.ijaresm.com

Comparison with Existing Studies

Our findings align with recent research on AI-driven stock market forecasting. Several studies post-2020 have
emphasized the superiority of deep learning techniques over conventional methods. For instance:

 Zhang et al. (2021) found that LSTMs outperformed ARIMA and SVM in stock price prediction due to their
ability to handle long-term dependencies.
 Chen et al. (2022) demonstrated that integrating financial news sentiment improved model accuracy by nearly
15% compared to models relying solely on historical price data.

Our study supports these conclusions, reinforcing the effectiveness of hybrid models that combine technical analysis,
deep learning, and market sentiment for more robust predictions.

Challenges and Limitations

Despite promising results, stock price prediction remains highly complex and uncertain due to several limitations:

1. Market Volatility – Machine learning models struggle to predict sudden market fluctuations caused by
unexpected events (e.g., financial crises, political instability, global pandemics).
2. Overfitting in Deep Learning Models – While LSTMs perform well, they can be prone to overfitting,
especially with limited training data. This issue requires careful hyperparameter tuning and regularization
techniques to improve generalization.
3. Data Quality and Availability – The accuracy of machine learning models depends heavily on high-quality,
real-time data. The lack of structured and updated financial data can affect prediction performance.
4. Computational Cost – Deep learning models, particularly LSTMs and GRUs, require significant
computational power, making real-time stock prediction challenging for individual traders and small firms.

Future Research Directions

To enhance stock price prediction, future research should focus on:

 Developing hybrid models that integrate fundamental analysis, reinforcement learning, and advanced deep
learning architectures.
 Improving real-time data processing by leveraging big data analytics and cloud computing for faster and more
efficient predictions.
 Exploring alternative financial indicators such as social media trends, investor sentiment, and blockchain-based
trading patterns.
 Enhancing model interpretability to make AI-driven predictions more transparent and reliable for investors and
financial analysts.

Conclusion & Future Scope

CONCLUSION

This research demonstrates the effectiveness of machine learning models in predicting stock prices using historical
data, technical indicators, and market sentiment analysis. Among the models analyzed, Long Short-Term Memory
(LSTM) networks exhibited the highest accuracy due to their ability to capture temporal dependencies and long-term
patterns in stock market trends. Traditional models like Linear Regression, Decision Trees, and Support Vector
Machines (SVM) performed moderately well but lacked the predictive power required for high-frequency trading and
volatile markets.

Additionally, integrating sentiment analysis from financial news and social media contributed to improved accuracy,
confirming that investor psychology and market sentiment play a critical role in stock price fluctuations. Feature
engineering techniques, such as moving averages, Bollinger Bands, and macroeconomic indicators, also proved
instrumental in enhancing model performance.

However, the study also highlights several challenges, including market unpredictability, overfitting in deep learning
models, and the requirement for high-quality real-time data. Despite these limitations, the results indicate that machine
learning provides a promising approach for financial forecasting, enabling traders and investors to make more data-
driven decisions.

Future Scope
Stock price prediction remains an evolving field with several areas for future research and improvement:

Page | 953
 International Journal of All Research Education and Scientific Methods (IJARESM),
ISSN: 2455-6211, Volume 13, Issue 5, May-2025, Available online at: www.ijaresm.com

Hybrid Models for Enhanced Accuracy

o Combining deep learning (LSTM, GRU) with reinforcement learning can enhance stock market forecasting by
enabling adaptive decision-making based on changing market conditions.
o Exploring ensemble learning techniques that integrate multiple models (e.g., Random Forest + LSTM) for
improved generalization.

Real-Time Data Integration

o Incorporating real-time financial news, economic reports, and social media analytics using natural language
processing (NLP) can further refine stock market predictions.
o Utilizing high-frequency trading (HFT) data to develop models capable of making near-instantaneous predictions.

Blockchain and Decentralized Finance (DeFi) Data

o Future research can explore the impact of cryptocurrency market trends and their correlation with traditional stock
markets.
o Leveraging blockchain-based financial records for more transparent and secure data-driven trading.

Improved Interpretability of AI Models

o Developing more interpretable and explainable AI (XAI) models to make financial predictions more transparent
and trustworthy for investors.
o Implementing attention mechanisms in deep learning models to understand which features contribute most to
stock price movements.

Reducing Overfitting and Improving Model Robustness

o Exploring regularization techniques, Bayesian optimization, and dropout strategies to prevent overfitting in deep
learning models.
o Using transfer learning to adapt pre-trained models from other financial markets for stock price forecasting.

Cross-Market and Global Stock Analysis

o Expanding the research to include multi-market predictions, analyzing the impact of global events on different
stock indices (e.g., S&P 500, NASDAQ, NIFTY 50).
o Developing models that account for geopolitical events, oil prices, and foreign exchange rates to enhance global
market prediction.

REFERENCES

[1]. Zhang, Y., Chen, K., & Li, X. (2021). Stock Market Prediction with LSTM: A Comparative Study on
Different Market Trends. Journal of Financial Analytics, 12(3), 45-60.
[2]. Chen, H., Wang, J., & Zhao, L. (2022). Integrating Sentiment Analysis and Machine Learning for Stock
Market Forecasting. International Journal of Data Science, 18(2), 134-152.
[3]. Gupta, R., & Sharma, P. (2023). Deep Learning-Based Hybrid Models for Stock Market Predictions: A
Review. IEEE Transactions on Computational Finance, 27(4), 287-302.
[4]. Singh, A., & Patel, J. (2021). Predicting Stock Market Volatility Using Neural Networks and Macroeconomic
Indicators. Finance and AI Journal, 9(1), 75-90.
[5]. Liu, X., Zhou, P., & Wang, T. (2022). A Hybrid CNN-LSTM Model for Stock Price Prediction.
Computational Economics, 39(5), 678-695.
[6]. Kumar, M., & Das, S. (2023). Machine Learning for High-Frequency Trading: Challenges and Opportunities.
Quantitative Finance Review, 15(2), 112-130.
[7]. Lee, J., & Park, S. (2021). Enhancing Stock Market Predictions with Explainable AI: A Case Study on Feature
Importance. Journal of Financial Machine Learning, 14(3), 201-217.
[8]. Tan, H., & Zhang, L. (2022). Financial News Sentiment and Stock Price Prediction: A Deep Learning
Approach. Artificial Intelligence in Finance, 11(4), 310-325.
[9]. Wang, Y., & Li, J. (2023). Reinforcement Learning for Stock Trading Strategies: A Review and Case Study.
Journal of Economic Computation, 22(1), 55-70.
[10]. Chakraborty, S., & Banerjee, R. (2021). Impact of Macroeconomic Indicators on Stock Market Prediction
Models. Global Finance Journal, 29(2), 98-115.

Page | 954