Stock price prediction

At the computational point of view, a fuzzy system has a layered structure, similar to an artificial neural network (ANN) of the radial basis function type. ANN learning algorithms can be employed for optimization of parameters in a fuzzy system. This neuro-fuzzy modeling approach has preference to explain solutions over completely black-box models, such as ANN. In this paper, we implement the design of experiment (DOE) technique to identify the significant parameters in the design of adaptive neuro-fuzzy inference systems (ANFIS) for stock price prediction.

This study compares different machine learning models for time series forecasting in financial data analysis. Models including ARIMA, LSTM, and GRU are applied to predict stock price movements. We measure the accuracy and computational efficiency of each model on various datasets and discuss their strengths and weaknesses in financial forecasting contexts. The findings suggest that deep learning models show significant improvement in capturing complex temporal patterns over traditional methods.

Stock price prediction is a challenging task and a lot of research continues to happen in the area. Portfolio construction is a process of choosing a group of stocks and investing in them optimally to maximize the return by minimizing the risk. Beginning from the Markowitz 'Modern Portfolio Theory' a lot of advancement has happened in the area of building efficient portfolios. An investor can get the best benefit out of the stock market if he/she

Stock market offers platform where people buy and sell shares of publicly listed companies. Generally stock prices are quite volatile; hence predicting them is daunting task. There are still many researches going to develop more accuracy in stock price prediction. Portfolio construction refers to allocation of different sector stocks optimally to achieve maximum return by taking minimum risk. A good portfolio can help investor earn maximum profit by taking minimum risk. Beginning from Dow Jones Theory a lot of advancement has happened in area of building efficient portfolio. In this project we have tried to predict future value of few stocks from six important sectors of Indian economy and also built portfolio. As part of the project, our team has conducted a study of performance of various Time series, machine learning (regression and classification) and deep learning models in stock price prediction on selected stocks from the chosen six important sectors of the economy. As part of building an efficient portfolio we have studied multiple portfolio optimization theories beginning from MPT (Modern Portfolio theory). We have built minimum variance portfolio and optimal risk portfolio for all the six chosen sectors by using past five years' daily stock price as training data and have also conducted back testing to check the performance of the portfolio. We look forward to continue our study in area of stock price prediction and asset allocation and consider this project as first stepping stone.

Deep learning (DL) has emerged as a powerful subset of machine learning (ML) and artificial intelligence (AI), outperforming traditional ML methods, especially in handling unstructured and large datasets. Its impact spans across various domains, including speech recognition, healthcare, autonomous vehicles, cybersecurity, predictive analytics, and more. However, the complexity and dynamic nature of real-world problems present challenges in designing effective deep learning models. Consequently, several deep learning models have been developed to address different problems and applications. In this article, we conduct a comprehensive survey of various deep learning models, including Convolutional Neural Network (CNN), Recurrent Neural Network (RNN), Temporal Convolutional Networks (TCN), Transformer, Kolmogorov-Arnold networks (KAN), Generative Models, Deep Reinforcement Learning (DRL), and Deep Transfer Learning. We examine the structure, applications, benefits, and limitations of each model. Furthermore, we perform an analysis using three publicly available datasets: IMDB, ARAS, and Fruit-360. We compared the performance of six renowned deep learning models: CNN, RNN, Long Short-Term Memory (LSTM), Bidirectional LSTM, Gated Recurrent Unit (GRU), and Bidirectional GRU alongside two newer models, TCN and Transformer, using the IMDB and ARAS datasets. Additionally, we evaluated the performance of eight CNN-based models, including VGG (Visual Geometry Group), Inception, ResNet (Residual Network), InceptionResNet, Xception (Extreme Inception), MobileNet, DenseNet (Dense Convolutional Network), and NASNet (Neural Architecture Search Network), for image classification tasks using the Fruit-360 dataset.

The growing financial complexity causes people to lose control over their financial expenses and savings goals together with their long-term planning requirements. Most personal finance management tools available today do not deliver forecasting statistics which match specific financial behavior patterns of their users. This paper unveils WealthGuardian which represents a smart expense management system which uses machine learning (ML) methodology to predict forthcoming expenses while improving budgetary restrictions and producing custom insurance policy suggestions. Support Vector Machines (SVM) and Random Forest operate together as robust ML algorithms to examine historical spending patterns and income behavior of users through the proposed system. Through model training these systems generate real-time expense forecasts which provide users an advanced financial perspective insight. The system merges data from external banks through APIs together with insurance plan databases to implement automated secure data transmissions which yield timely useful insights.

This research focuses on predicting stock prices using Gated Recurrent Units (GRUs), a type of Recurrent Neural Network (RNN) that effectively captures sequential dependencies in time series data. The model leverages historical stock data and presents results using interactive dashboards in Power BI, enhancing decision-making and interpretability for stakeholders. The study evaluates the accuracy and efficiency of GRU-based models against traditional approaches, demonstrating improved forecasting capabilities. Additionally, the research explores the impact of various hyper-parameters on model performance and compares GRU with other deep learning architectures, such as LSTMs, to determine the optimal approach for financial time series forecasting. The system's ability to detect trends, mitigate noise, and provide real-time insights makes it a valuable tool for investors and financial analysts. Furthermore, this study examines real-world applications, industry adoption, and the scalability of GRU-based predictive models in financial markets, ensuring robust performance across varying market conditions.

Accurate prediction of the financial markets can provide many benefits, of which underlying economic stability is probably the most important. This area has understandably attracted a significant amount of interest from the research community, and has inspired a diverse range of approaches with varying degrees of success. Gold is a particular commodity which has attracted considerable attention since it was first smelted for ornaments and jewellery by the Egyptians in 3600BC. In uncertain economic times it is regularly used as a safe-haven commodity, and is why there is considerable attention given to enhancing the accuracy of gold prices prediction methods. Previous attempts at gold price prediction have used a variety of econometric and machine learning techniques. In particular Long Short-Term Networks (LSTMs) and more recently an ensemble of Convolutional Neural Networks (CNNs) and LSTMs have been found to have had considerable level of success in time series prediction. In this...

The market state changes when a new piece of information arrives. It affects decisions made by investors and is considered to be an important data source that can be used for financial forecasting. Recently information derived from news articles has become a part of financial predictive systems. The usage of news articles and their forecasting potential have been extensively researched. However, so far no attempts have been made to utilise different categories of news articles simultaneously. This paper studies how the concurrent, and appropriately weighted, usage of news articles, having different degrees of relevance to the target stock, can improve the performance of financial forecasting and support the decision-making process of investors and traders. Stock price movements are predicted using the multiple kernel learning technique which integrates information extracted from multiple news categories while separate kernels are utilised to analyse each category. News articles are partitioned according to their relevance to the target stock, its sub industry, industry, group industry and sector. The experiments are run on stocks from the Health Care sector and show that increasing the number of relevant news categories used as data sources for financial forecasting improves the performance of the predictive system in comparison with approaches based on a lower number of categories.

The market state changes when a new piece of information arrives. It affects decisions made by investors and is considered to be an important data source that can be used for financial forecasting. Recently information derived from news articles has become a part of financial predictive systems. The usage of news articles and their forecasting potential have been extensively researched. However, so far no attempts have been made to utilise different categories of news articles simultaneously. This paper studies how the concurrent, and appropriately weighted, usage of news articles, having different degrees of relevance to the target stock, can improve the performance of financial forecasting and support the decision-making process of investors and traders. Stock price movements are predicted using the multiple kernel learning technique which integrates information extracted from multiple news categories while separate kernels are utilised to analyse each category. News articles are partitioned according to their relevance to the target stock, its sub industry, industry, group industry and sector. The experiments are run on stocks from the Health Care sector and show that increasing the number of relevant news categories used as data sources for financial forecasting improves the performance of the predictive system in comparison with approaches based on a lower number of categories.

Accurately forecasting financial time-series data is a challenging task due to the dynamic and volatile nature of stock markets. This study introduces a hybrid Temporal Convolutional Network (TCN) and Long Short-Term Memory (LSTM) model designed to improve stock price forecasting for the Muscat Securities Market (MSM). Unlike standalone deep learning models, this hybrid approach effectively captures both short-term and long-term dependencies, leading to improved predictive accuracy. Trained on 24 years of historical MSM data (2000-2024), the model was evaluated using Mean Absolute Error (MAE) and Root Mean Squared Error (RMSE). The hybrid model outperformed both standalone architectures, achieving the lowest MAE (206.29) and RMSE (314.31). This research advances financial forecasting by introducing a hybrid TCN-LSTM model specifically optimized for the Muscat Securities Market (MSM), a relatively underexplored financial domain. The study bridges the gap in existing models by enhancing predictive performance through an innovative fusion of deep learning techniques. The study contributes to financial forecasting research by demonstrating how hybrid deep learning models can enhance market prediction accuracy, providing valuable insights for investors and financial analysts. Future research directions include the integration of adaptive learning mechanisms and external financial indicators for further performance enhancement.

Machine learning (ML) techniques have emerged as promising tools for enhancing market forecasting compared to traditional methods. This research conducts a systematic literature review to delineate current trends and future trajectories in ML-driven stock market prediction studies. Peer-reviewed journal articles spanning the last twenty years were classified into four primary categories: artificial neural networks, support vector machines, genetic algorithms in conjunction with other methods, and hybrid or alternative AI approaches. Each category underwent scrutiny to unveil commonalities, distinctive perspectives, constraints, and areas necessitating further exploration. The outcomes furnish valuable insights and suggestions for forthcoming research endeavours in this field.

We introduce MinimalRNN, a new recurrent neural network architecture that achieves comparable performance as the popular gated RNNs with a simplified structure. It employs minimal updates within RNN, which not only leads to efficient learning and testing but more importantly better interpretability and trainability. We demonstrate that by endorsing the more restrictive update rule, MinimalRNN learns disentangled RNN states. We further examine the learning dynamics of different RNN structures using input-output Jacobians, and show that MinimalRNN is able to capture longer range dependencies than existing RNN architectures.

Long term prediction of future prices and returns of financial time series, particularly asset classes such as indices and stocks, has been one of the most challenging problems in both academia and finance. There are many contributing factors to the volatility of stocks and indices prices, such as macroeconomic news, trader psychology, algorithmic trading, and natural disasters such as pandemics, which leads to noisy and non-stationary time series. Although classical statistical prediction models such as ARIMA can predict the next time lag of time series accurately, their performance sharply falls when it comes to predicting longer periods. Besides, they can not generalize beyond a single share or market. As the size, diversity and frequency of financial data evolve, and massive computation power being easily accessible nowadays, we need modern approaches such as machine learning and deep learning to handle the issues of predicting financial time series. LSTM is a variety of in-depth knowledge that has achieved tremendous successes in many fields, such as sounds and speech recognition. LSTMs are also capable of learning from sequence data and addressing problems of RNNs such as vanishing and exploding gradients. This study proposes two LSTM models to predict the close prices and returns of the S&P 500 index and three stocks from the S&P5oo index, including Apple, HP, and Walmart. These models are compared to each other in terms of prediction accuracy with RMSE as the performance measure. Additionally, the prediction accuracies of both LSTM models are compared to the ARIMA(5,1,0) model in predicting close prices. iv Subsequently, this study identifies a proper Cross-validation method particular to financial time series, to address the training data leakage problem Furthermore, this study identifies the most impactful model hyperparameters in the training process. The results of the experiment show the Stacked LSTM model, which incorporates a convolution layer, outperforms the simple LSTM model in predicting both closing prices and returns. Moreover, the result indicates that both LSTM models have achieved a considerably better result in comparison to the ARIMA model. The result of this research affirms that LSTM models can be utilized to predict the future prices of stocks and indices effectively predict future close prices and returns of stocks and indices. The LSTM models trained in this research can be used as software components in various financial applications such as portfolio allocation, risk assessment and algorithmic trading.

Myotonia, which refers to delayed muscle relaxation after contraction, is the main symptom of myotonic dystrophy patients. We propose a hierarchical attention-based temporal convolutional network (HA-TCN) for myotonic dystrohpy diagnosis from handgrip time series data, and introduce mechanisms that enable model explainability. We compare the performance of the HA-TCN model against that of benchmark TCN models, LSTM models with and without attention mechanisms, and SVM approaches with handcrafted features. In terms of classification accuracy and F1 score, we found all deep learning models have similar levels of performance, and they all outperform SVM. Further, the HA-TCN model outperforms its TCN counterpart with regards to computational efficiency regardless of network depth, and in terms of performance particularly when the number of hidden layers is small. Lastly, HA-TCN models can consistently identify relevant time series segments in the relaxation phase of the handgrip time se...

This study investigates stock price prediction using LSTM, a type of recurrent neural network. By analyzing Tesla's stock data (2015–2022), the model achieved notable accuracy with an R2 score of 0.92, showcasing LSTM's potential in handling complex time-series data for financial forecasting.

S&P 500 is the largest and state-of-the-art stock market index in North America, which attracted a wide range of audience. The primary objective of this study is to compare the widely used four stock forecasting approaches: Long Term- Short Term Memory (LSTM), Gated Recurrent Unit, (GRU), Convolutional Neural Network, (CNN) and traditional
forecasting approach: Auto- Regressive Integrated Moving Average (ARIMA) to identify the best and more robust forecasting model for daily and weekly closing price on the S&P 500 financial sector. Thus, we developed and compared the performance and quality of these AI based approaches with baseline traditional ARIMA model using well- defined two statistical metrics, Root Mean Square Error (RMSE) and Mean Absolute Error (MAE) as the evaluation criterion. In the scope of our study, we have found that the LSTM outperforms (with more than 15% improvement in RMSE and with more than 30% improvement in MAE compared to
ARIMA) 2 out of 3 train/test data splits compared to other proposed deep learning approaches including GRU and traditional ARIMA models with respect to two widely used RMSE and MAE evaluation metrics for daily closing price forecasting in the S&P 500 Financial Sector. Additionally, in the weekly closing price forecasting models, the traditional ARIMA
model outperforms all deep learning models on 2 out of 3 train/test data splits with respect to the statistical metric RMSE

In this study, we propose a deep neural generative model for predicting daily stock price movements given news articles. Approaches involving conventional technical analysis have been investigated to identify certain patterns in past price movements, which in turn helps to predict future price movements. However, the financial market is highly sensitive to specific events, including corporate buyouts, product releases, and the like. Therefore, recent research has focused on modeling relationships between these events that appear in the news articles and future price movements; however, a very large number of news articles are published daily, each article containing rich information, which results in overfitting to past price movements used for parameter adjustment. Given the above, we propose a model based on a generative model of news articles that includes price movement as a condition, thereby avoiding excessive overfitting thanks to the nature of the generative model. We evaluate our proposed model using historical price movements of Nikkei 225 and Standard & Poor's 500 Stock Index, confirming that our model predicts future price movements better than such conventional classifiers as support vector machines and multilayer perceptrons. Further, our proposed model extracts significant words from news articles that are directly related to future stock price movements.

In the stock market, which is a dynamic, complex, nonlinear and non-parametric environment, accurate prediction is crucial for trading strategy. It is assumed that news articles affect the stock market. We investigated the relationship[1] between headline’s sentiment of news and their impact on stock prices changes. To show this relationship, we applied the sentiment data and the price difference between the day before the news was published and the day of the news, to machine learning regression and classification models. Regression is used to predict changes and classification is used to decide whether to buy or sell stocks.  We used three stock datasets named Apple, Amazon and AXP and the results are shown in the mentioned dataset that using news with negative sentiments can make predictions just as correctly as using news with both positive and negative sentiments. In regression and classification models, Random Forest outperformed other machine learning algorithms in predicting stock price changes using news sentiment analysis. Additionally, we depicted that the results of computer and human tagging were almost similar, showing that using computer tools for text tagging will allow to tag text much more quickly and easily.

One of the most intricate machine learning problems is the share value prediction. Stock market prediction is an activity in which investors need fast and accurate information to make effective decisions. Moreover, the behavior of stock prices is uncertain and hard to predict. For these reasons, stock price prediction is an important process and a challenging one. This leads to the research of finding the most effective prediction model that generates the most accurate prediction with the lowest error percentage. Prices of stocks are depicted by time series data and neural networks are trained to learn the patterns from trends in the existing data. This system employed algorithm using LSTM to improve the accuracy of stock price prediction.

Stock price prediction is challenging due to global economic instability, high volatility, and the complexity of financial markets. Hence, this study compared several machine learning algorithms for stock market prediction and further examined the influence of a sentiment analysis indicator on the prediction of stock prices. Our results were twofold. Firstly, we used a lexiconbased sentiment analysis approach to identify sentiment features, thus evidencing the correlation between the sentiment indicator and stock price movement. Secondly, we proposed the use of GRUvader, an optimal gated recurrent unit network, for stock market prediction. Our findings suggest that stand-alone models struggled compared with AI-enhanced models. Thus, our paper makes further recommendations on latter systems.

ABSTRACT
This research work focuses at investigating the viability of using deep learning techniques for stock market
prediction in order to establish the role that time series plays in the area of financial prediction. Markets by their
nature involve a great deal of variability and are contingent upon a number of factors such as Economic
globalization, social factors, and charts trends. Earlier, simpler models such as ARIMA and linear regression
techniques have been used, however, due to their inability to model complex patterns that are non-linear in shape
the more complex machine learning algorithms have been integrated. RNN and LSTM categories are effective in
analyzing sequential data and hence can be used when predicting the stock market. This research aims at
understanding the efficiency of these models in predicting stock prices given that this study will use several
financial datasets among them the previous stock prices and trading volumes. Performance indicators including
correct percentage, mean absolute error, and root mean square error are employed on the models. The outcomes
indicate that the deep learning models generate more accurate predictions of the future stock price changes than
traditional methods because of their inherent long-term dependency modeling mechanism and shared mechanism.
Moreover, the study shows that the issue of feature engineering and hyperparameter tuning should be given a more
comprehensive approach due to its significance in increasing model accuracy. The studies imply that embedded the
deep learning approach, it is possible to improve the degree of financial forecasting and enhance investors and
analysts skills in decision making. Future work will be directed toward expanding the presented model for
increasing its robustness together with the usage of other features, for example, news sentiment and shrinkage data.
Keywords: Deep learning, stock market prediction, time-series data, LSTM, financial forecasting.

We propose a novel architecture of recurrent neural networks (RNNs) for causal prediction which we call Entangled RNN (E-RNN). To issue causal predictions, E-RNN can propagate the backward hidden states of Bi-RNN through an additional forward hidden layer. Unlike a 2-layer RNNs, all the hidden states of E-RNN depend on all the inputs seen so far. Furthermore, unlike a Bi-RNN, for causal prediction, E-RNN depends on both the forward and backward hidden states. Importantly, E-RNN is a general architecture that can be combined with various RNN techniques such as multi-layer, dropout, and GRU. Using three real-world datasets, we show that E-RNN significantly and consistently improves the performance of previous RNN architectures with the same complexity.

A comparative analysis of deep learning models and traditional statistical methods for stock price prediction uses data from the Nigerian stock exchange. Historical data, including daily prices and trading volumes, are employed to implement models such as Long Short Term Memory (LSTM) networks, Gated Recurrent Units (GRUs), Autoregressive Integrated Moving Average (ARIMA), and Autoregressive Moving Average (ARMA). These models are assessed over three-time horizons: short-term (1 year), medium-term (2.5 years), and long-term (5 years), with performance measured by Mean Squared Error (MSE) and Mean Absolute Error (MAE). The stability of the time series is tested using the Augmented Dickey-Fuller (ADF) test. Results reveal that deep learning models, particularly LSTM, outperform traditional methods by capturing complex, nonlinear patterns in the data, resulting in more accurate predictions. However, these models require greater computational resources and offer less interpretability than traditional approaches. The findings highlight the potential of deep learning for improving financial forecasting and investment strategies. Future research could incorporate external factors such as social media sentiment and economic indicators, refine model architectures, and explore real-time applications to enhance prediction accuracy and scalability.

A comparative analysis of deep learning models and traditional statistical methods for stock price prediction uses data from the Nigerian stock exchange. Historical data, including daily prices and trading volumes, are employed to implement models such as Long Short Term Memory (LSTM) networks, Gated Recurrent Units (GRUs), Autoregressive Integrated Moving Average (ARIMA), and Autoregressive Moving Average (ARMA). These models are assessed over three-time horizons: short-term (1 year), medium-term (2.5 years), and long-term (5 years), with performance measured by Mean Squared Error (MSE) and Mean Absolute Error (MAE). The stability of the time series is tested using the Augmented Dickey-Fuller (ADF) test. Results reveal that deep learning models, particularly LSTM, outperform traditional methods by capturing complex, nonlinear patterns in the data, resulting in more accurate predictions. However, these models require greater computational resources and offer less interpretability than traditional approaches. The findings highlight the potential of deep learning for improving financial forecasting and investment strategies. Future research could incorporate external factors such as social media sentiment and economic indicators, refine model architectures, and explore real-time applications to enhance prediction accuracy and scalability.

Deep learning, one of the most remarkable techniques of machine learning, has been a major success in many fields, including image processing, speech recognition, and text understanding. It is powerful engines capable of learning arbitrary mapping functions, not require a scaled or stationary time series as input, support multivariate inputs, and support multi-step outputs. All of these features together make deep learning useful tools when dealing with more complex time series prediction problems involving large amounts of data, and multiple variables with complex relationships. This paper provides an overview of the most common Deep Learning types for time series forecasting, Explain the relationships between deep learning models and classical approaches to time series forecasting. A brief background of the particular challenges presents in time-series data and the most common deep learning techniques that are often used for time series forecasting is provided. Previous studies that applied deep learning to time series are reviewed.

Deep learning, one of the most remarkable techniques of machine learning, has been a major success in many fields, including image processing, speech recognition, and text understanding. It is powerful engines capable of learning arbitrary mapping functions, not require a scaled or stationary time series as input, support multivariate inputs, and support multi-step outputs. All of these features together make deep learning useful tools when dealing with more complex time series prediction problems involving large amounts of data, and multiple variables with complex relationships. This paper provides an overview of the most common Deep Learning types for time series forecasting, Explain the relationships between deep learning models and classical approaches to time series forecasting. A brief background of the particular challenges presents in time-series data and the most common deep learning techniques that are often used for time series forecasting is provided. Previous studies that applied deep learning to time series are reviewed.

Egypt’s agricultural sector plays a critical role in the country’s economy, with wheat cultivation being vital for ensuring food
security. However, the challenges faced by wheat farming in Egypt, such as climate change, water scarcity, and pest infestations,
contribute to yield fluctuations, highlighting the need for accurate and timely predictions of wheat productivity. To address
this need, time series analysis is employed, which involves analyzing data collected at regular intervals over a specific time
frame. Time series data can encompass multiple variables recorded simultaneously at each interval, known as multivariate time
series data. Traditional statistical methods have been commonly used for time series analysis. However, these methods have
limitations when dealing with nonlinear and complex data patterns, especially in agricultural data exhibiting spatiotemporal
characteristics. Deep learning techniques have emerged as a promising solution to address these limitations. In this paper, we
have developed the multivariate multi-head attention temporal convolutional networks (MATCN) model specifically to handle
the spatiotemporal nature of multivariate time series data in the context of wheat crop productivity prediction in Egypt. The
experimental results revealed that our proposed MATCN model exhibited the most effective prediction performance when
compared to other state-of-the-art models, achieving a mean absolute percentage error (MAPE) values ranging from 0.091 to
5.5%.

As the integration of robots into various industries becomes increasingly prevalent, understanding the nuanced effects on the workforce is imperative.This study investigates the implications of Human-Robot Collaboration (HRC) in the workplace, specifically focusing on its impact on employee wellbeing and productivity. Leveraging advanced deep learning methodologies, the proposed Long Short-Term Memory-Gated Recurrent Unit (LSTM-GRU) model is introduced as a novel approach for assessing these critical factors. Implemented in Python, the LSTM-GRU model serves as a powerful tool for assessing the multifaceted dynamics of HRC. The utilization of deep learning techniques, coupled with the LSTM-GRU architecture, allows for a more nuanced understanding of the intricate relationships between human and robotic collaborators in the workplace. The findings of this study contribute valuable insights for organizations seeking to optimize HRC integration, highlighting the significance of advanced computational approaches in gauging the impact on employee well-being and productivity. The Python implementation further enhances the model's accessibility, enabling researchers and practitioners to adopt and adapt this methodology in diverse workplace settings. The study employs a diverse set of performance metrics, including accuracy, precision, recall, and F1-Score, to comprehensively evaluate the effectiveness of the proposed model. The results reveal a remarkable accuracy of 99.11% for the LSTM-GRU model, showcasing its superior ability to capture nuanced temporal dependencies in comparison to existing methods such as Convolutional Neural Network(CNN), CNN-LSTM, and Bidirectional Long Short-Term Memory (BiLSTM). The proposed model's accuracy surpasses these methods by 3.23%, underscoring its robustness in accurately identifying and predicting aspects related to employee well-being and productivity.

At the computational point of view, a fuzzy system has a layered structure, similar to an artificial neural network (ANN) of the radial basis function type. ANN learning algorithms can be employed for optimization of parameters in a fuzzy system. This neuro-fuzzy modeling approach has preference to explain solutions over completely black-box models, such as ANN. In this paper, we implement the design of experiment (DOE) technique to identify the significant parameters in the design of adaptive neuro-fuzzy inference systems (ANFIS) for stock price prediction.

Financial Forecast is an estimate of future financial outcomes for a company. A financial forecast is an economist's best guess of what will happen to a company in financial terms over a given time period. Data mining has made a significant role in the prediction techniques for financial applications. Technology such as data ware house and data mining has made a significant contribution for the prediction in the entire service sector of the organization .The Data mining techniques are used to extract hidden patterns, from large amount of financial data with the effective usage of sentiment features and opinion mining. The paper describes the effective usage of data mining predictive techniques for the financial applications.

Nowadays, the information obtainable from the markets are potentially limitless. Economic theory has always supported the possible advantage obtainable from having more information than competitors, however quantifying the advantage that these can give has always been a problem. In particular, in this paper we study the amount of information obtainable from the markets taking into account only the time series of the prices, through the use of a specific Generative Adversarial Network. We consider two types of financial instruments traded on the market, stocks and cryptocurrencies: the first are traded in a market subject to opening and closing hours, whereas cryptocurrencies are traded in a 24/7 market. Our goal is to use this GAN to be able to “convert” the amount of information that the different instruments can have in discriminative and predictive power, useful to improve forecast. Finally, we demonstrate that by using the initial dataset with the 5 most important feature useds ...

In the dynamic landscape of financial markets, accurately predicting stock price movements is of paramount importance for investors, traders, and financial institutions. This second paper in the series delves into the pivotal role of deep learning architectures, specifically recurrent neural networks (RNNs), long short-term memory networks (LSTMs), and gated recurrent units (GRUs), in this critical task. Through a comparative analysis, we evaluate the efficacy and suitability of these architectures for stock price forecasting. Leveraging real-world financial data, we scrutinize their accuracy, robustness, and computational efficiency. Our findings provide valuable insights into the application of advanced deep learning methodologies in financial forecasting, empowering stakeholders with actionable intelligence for navigating the complexities of the financial markets.

In the dynamic landscape of financial markets, accurately predicting stock price movements is of paramount importance for investors, traders, and financial institutions. This second paper in the series delves into the pivotal role of deep learning architectures, specifically recurrent neural networks (RNNs), long short-term memory networks (LSTMs), and gated recurrent units (GRUs), in this critical task. Through a comparative analysis, we evaluate the efficacy and suitability of these architectures for stock price forecasting. Leveraging real-world financial data, we scrutinize their accuracy, robustness, and computational efficiency. Our findings provide valuable insights into the application of advanced deep learning methodologies in financial forecasting, empowering stakeholders with actionable intelligence for navigating the complexities of the financial markets.

Stock prices in each industry are one of the major issues in the stock market. Given the increasing number of shareholders in the stock market and their attention to the price of different stocks in transactions, the prediction of the stock price trend has become significant. Many people use the share price movement process when com-paring different stocks while investing, and also want to predict this trend to see if the trend continues to increase or decrease over time. In this research, stock price prediction for 1170 years -company during 2011-2016 (a six-year period) of listed companies in stock exchange has been studied using the machine learning method (Chaid rule-based algorithm and Particle Swarm Optimization Algorithm). The results of the research show that there is a significant relationship between earnings per share, e / p ratio, company size, inventory turnover ratio, and stock returns with stock prices. Also, particle swarm optimization (pso) algorithm has a good abil...

Over the last few years, networks' infrastructures are experiencing a profound change initiated by Software Defined Networking (SDN) and Network Function Virtualization (NFV). In such networks, avoiding the risk of service degradation increasingly involves predicting the evolution of metrics impacting the Quality of Service (QoS), in order to implement appropriate preventive actions. Recurrent neural networks, in particular Long Short Term Memory (LSTM) networks, already demonstrated their efficiency in predicting time series, in particular in networking, thanks to their ability to memorize long sequences of data. In this paper, we propose an improvement that increases their accuracy by combining them with filters, especially the Fast Fourier Transform (FFT), in order to better extract the characteristics of the time series to be predicted. The proposed approach allows improving prediction performance significantly, while presenting an extremely low computational complexity at run-time compared to classical techniques such as Auto-Regressive Integrated Moving Average (ARIMA), which requires costly online operations.

1 | Introduction Various tmethods for predicting stock prices include technical analysis, mathematical and statistical approaches, and econometric models. The primary motivation behind the stock market prediction is rooted in the Efficient Market Hypothesis (EMH) and the pursuit of financial gains. EMH is of particular significance because it serves as a foundational concept underpinning the effectiveness of prediction. An efficient market implies that all available information is reflected in stock prices. Conversely, an inefficient market suggests the presence of individuals and entities with privileged access to additional market information. It underscores the idea that information can significantly impact the decisions made by investors, consequently leading to changes in stock prices [1].

In this study, it is aimed to compare the performances of the algorithms by predicting the movement directions of stock market indexes in developed countries by employing machine learning algorithms (MLMs) and determining the best estimation algorithm. For this purpose, the movement directions of indexes such as the NYSE 100 (the USA), NIKKEI 225 (Japan), FTSE 100 (the UK), CAC 40 (France), DAX 30 (Germany), FTSE MIB (Italy), and TSX (Canada) were estimated by employing the decision tree, random forest k-nearest neighbor, naive Bayes, logistic regression, support vector machines and artificial neural network algorithms. According to the results obtained, artificial neural networks were found to be the best algorithm for NYSE 100, FTSE 100, DAX 30 and FTSE MIB indices, while logistic regression was determined to be the best algorithm for the NIKKEI 225, CAC 40, and TSX indices. The artificial neural networks, which exhibited the highest average prediction performance, have been determined as the best prediction algorithm for the stock market indices of developed countries. It was also noted that artificial neural networks, logistic regression, and support vector machines algorithms were capable of predicting the directional movements of all indices with an accuracy rate of over 70 %.

Predicting the Stock movement attracts much attention from both industry and academia. Despite such significant efforts, the results remain unsatisfactory due to the inherently complicated nature of the stock market driven by factors including supply and demand, the state of the economy, the political climate, and even irrational human behavior. Recently, Generative Adversarial Networks (GAN) have been extended for time series data; however, robust methods are primarily for synthetic series generation, which fall short for appropriate stock prediction. This is because existing GANs for stock applications suffer from mode collapse and only consider one-step prediction, thus underutilizing the potential of GAN. Furthermore, merging news and market volatility are neglected in current GANs. To address these issues, we exploit expert domain knowledge in finance and, for the first time, attempt to formulate stock movement prediction into a Wasserstein GAN framework for multi-step prediction. We propose IndexGAN, which includes deliberate designs for the inherent characteristics of the stock market, leverages news context learning to thoroughly investigate textual information and develop an attentive seq2seq learning network that captures the temporal dependency among stock prices, news, and market sentiment. We also utilize the critic to approximate the Wasserstein distance between actual and predicted sequences and develop a rolling strategy for deployment that mitigates noise from the financial market. Extensive experiments are conducted on real-world broad-based indices, demonstrating the superior performance of our architecture over other state-of-the-art baselines, also validating all its contributing components.

Proposing a new precise price forecasting method is still a challenging task as electricity price signals generally exhibit various complex features. In this paper, a new approach for electricity price forecasting called hybrid local general regression neural network, and harmony search algorithm (LGRNN-HSA) is proposed. The proposed LGRNN-HSA is developed by combining the coordinate delay (CD) method, local forecasting paradigm, general regression neural network (GRNN) and harmony search algorithm (HSA). The CD is employed in the proposed method to reconstruct the time series dataset. Then the local forecasting paradigm is utilized with GRNN to predict the future price based on the nearest neighbours only so that the shortcomings in global forecasting methods can be overcome. To enhance the forecasting accuracy, the HSA is employed to optimize not only the parameters of local forecasting paradigm but also the smooth parameter which has a great effect on the accuracy of GRNN. In LGRNN-HSA, a different smooth parameter for every training point is utilized instead of utilizing a constant value in GRNN. To verify the forecasting accuracy of the LGRNN-HSA, a real-world price and demand dataset is employed. The simulation results prove that the LGRNN-HSA significantly improved forecasting accuracy compared to other methods.

This dissertation is a very important milestone for me. However, this is not an individual journey, it is the cumulative work of many people who have contributed to my life and to my master's adventure. I am truly grateful to all who have become a part of this incredible journey. I am very grateful to my supervisor Professor Mauro Castelli who has always shown availability and immediate support to all my questions and doubts along this project. It was very important since it allowed me to continue all the work and to achieve my goals towards the end of this academic path. Since this has been a long journey, my family and friends have also a crucial importance. I want to express my thanks to my parents, Paulo and Cristina, who have taught me about determination and persistence. Also, my sister Melissa had an important role during this phase since she had always been here reminding me that her dissertation was almost done and mine should be in the end too so that we could celebrate it together. A special thanks to Marta and David who have been present all the way since the beginning. All their patience, encouragement and friendship were very important to me. I cannot emphasize enough my deep gratitude for their role and support over this last year.

Financial time-series prediction has been long and the most challenging issues in financial market analysis.The deep neural networks is one of the excellent data mining approach has received great attention by researchers in several areas of time-series prediction since last 10 years. Convolutional neural network (CNN) and recurrent neural network (RNN) models have become the mainstream methods for financial predictions. In this paper, we proposed to combine architectures, which exploit the advantages of CNN and RNN simultaneously, for the prediction of trading signals. Our model is essentially presented to financial time series predicting signals through a CNN layer, and directly fed into a gated recurrent unit (GRU) layer to capture long-term signals dependencies. GRU model perform better in sequential learning tasks and solve the vanishing gradients and exploding issue in standard RNNs. We evaluate our model on three datasets for stock indexes of the Hang Seng Indexes (HSI), the ...

In this study, financial prediction models have been developed over the silver / ounce parity using deep learning architectures. LSTM and ARIMA architectures, which are deep learning algorithms, are used. By loading the training and test data into the established algorithms, the system was learned and a graphical estimation was requested on the silver / ounce parity for the next 10 days. Written algorithms can produce different results each time they are run. However, in the graphs we have taken as an example, the graph created with the ARIMA architecture has produced a more realistic result by specifying a range and making an upward forecast. The prediction chart we obtained with the LSTM architecture did not create a much decrease or upward forecast. However, as a feature of the LSTM algorithm, it clearly predicted the daily closing values, and did not specify an estimation as a range and direction as in the study with the ARIMA architecture. It should not be forgotten that these algorithms are dynamic and can give different results in predictions even when they are run with the same data. According to the results obtained in the research, although the LSTM architecture clearly stated the daily closing values as numbers, the estimation study made with the ARIMA architecture produced a result closer to the graph in terms of both interval and direction.

Financial time series prediction has been a key topic of interest among researchers considering the complexity of the domain and also due to its significant impact on a wide range of applications. In contrast to one-step ahead prediction, multi-step forecasting is more desirable in the industry but the task is more challenging. In recent days, advancement in deep learning has shown impressive accomplishments across various tasks including sequence learning and time series forecasting. Although most previous studies are focused on applications of deep learning models for single-step ahead prediction, multi-step financial time series forecasting has not been explored exhaustively. This paper aims at extensively evaluating the performance of various state-of-the-art deep learning models for multiple multi-steps ahead prediction horizons on real-world stock and forex markets dataset. Specifically, we focus on Long-Short Term Memory (LSTM) network and its variations, Encoder-Decoder based sequence to sequence models, Temporal Convolution Network (TCN), hybrid Exponential Smoothing-Recurrent Neural Networks (ES-RNN) and Neural Basis Expansion Analysis for interpretable Time Series forecasting (N-BEATS). Experimental results show that the latest deep learning models such as N-BEATS, ES-LSTM and TCN produced better results for all stock market related datasets by obtaining around 50% less Root Mean Squared Error (RMSE) and Mean Absolute Error (MAE) scores for each prediction horizon as compared to other models. However, the conventional LSTM-based models still prove to be dominant in the forex domain by comparatively achieving around 2% less error values.

Over the years, researchers have strived to develop reliable and accurate predictive models for stock price prediction. The literature suggests that well-designed and refined predictive models can provide painstakingly precise estimates of future stock values. This project aims to showcase a comprehensive set of models for predicting stock prices, including time series, econometric, statistical, and machine learning-based approaches. The dataset includes ten industry leaders from different sectors and NIFTY50, spanning from January 2017 to December 2022. The models' performance was evaluated to determine which approach performs best for different sectors. The time series models employed include Holt's Linear Trend and Holt-Winters Exponential Smoothing, while the econometric model utilized is ARIMA. Additionally, the statistical model adopted is OLS, while several machine learning and deep learning models incorporated a range of techniques such as Random Forest, KNN, CNN, LSTM, etc. Besides predicting stock prices, the models were also designed to determine the direction of stock price movements. This project aims to provide insights into the best methods for predicting stock prices and direction across different industries by combining different models and approaches.

Over the years, researchers have strived to develop reliable and accurate predictive models for stock price prediction. The literature suggests that well-designed and refined predictive models can provide painstakingly precise estimates of future stock values. This project aims to showcase a comprehensive set of models for predicting stock prices, including time series, econometric, statistical, and machine learning-based approaches. The dataset includes ten industry leaders from different sectors and NIFTY50, spanning from January 2017 to December 2022. The models' performance was evaluated to determine which approach performs best for different sectors. The time series models employed include Holt's Linear Trend and Holt-Winters Exponential Smoothing, while the econometric model utilized is ARIMA. Additionally, the statistical model adopted is OLS, while several machine learning and deep learning models incorporated a range of techniques such as Random Forest, KNN, CNN, LS...

Predictive model design for accurately predicting future stock prices has always been considered an interesting and challenging research problem. The task becomes complex due to the volatile and stochastic nature of the stock prices in the real world which is affected by numerous controllable and uncontrollable variables. This paper presents an optimized predictive model built on longand-short-term memory (LSTM) architecture for automatically extracting past stock prices from the web over a specified time interval and predicting their future prices for a specified forecast horizon, and forecasts the future stock prices. The model is deployed for making buy and sell transactions based on its predicted results for 70 important stocks from seven different sectors listed in the National Stock Exchange (NSE) of India. The profitability of each sector is derived based on the total profit yielded by the stocks in that sector over a period from Jan 1, 2010 to Aug 26, 2021. The sectors are compared based on their profitability values. The prediction accuracy of the model is also evaluated for each sector. The results indicate that the model is highly accurate in predicting future stock prices.

Accurate time series forecasting is a fundamental challenge in data science, as it is often affected by external covariates such as weather or human intervention, which in many applications, may be predicted with reasonable accuracy. We refer to them as predicted future covariates. However, existing methods that attempt to predict time series in an iterative manner with auto-regressive models end up with exponential error accumulations. Other strategies that consider the past and future in the encoder and decoder respectively limit themselves by dealing with the past and future data separately. To address these limitations, a novel feature representation strategy-shifting-is proposed to fuse the past data and future covariates such that their interactions can be considered. To extract complex dynamics in time series, we develop a parallel deep learning framework composed of RNN and CNN, both of which are used in a hierarchical fashion. We also utilize the skip connection technique to improve the model's performance. Extensive experiments on three datasets reveal the effectiveness of our method. Finally, we demonstrate the model interpretability using the Grad-CAM algorithm.

This study applies neural architecture search (NAS) techniques to the modeling of highdimensional time series data such as multi-variate stock indices. It is known that traditional NAS method applies fully connected directed acyclic graph (DAG) for searching cell structures that requires high computational cost and cannot include the two-input operations such as the Hadamard product. To address the drawback of the DAG backbone, a novel two-input backbone cell architecture for recurrent neural networks is proposed, in which each candidate operation is also carried out with two inputs. Instead of using DAG, we simplify the backbone by considering the prior knowledge as an effective backbone such as preserving identity mappings. The cell structures will be incorporated in different types of model architectures including stacked long short-term memory (LSTM), gated recurrent unit (GRU) and attention-based encoder-decoder models. The experimental results on BRICS, G7 and G20 indices indicate that models with recurrent neural network (RNN) cells searched by the proposed backbone structure can significantly outperform baseline models including autoregressive integrated moving average model (ARIMA), vector autoregression (VAR), and stacked LSTM/GRUs. For neural architecture search, the proposed backbone is shown to be more effective compared to the classic differentiable architecture search (DARTS) in both uni-variate and multivariate time series prediction tasks. Further analysis demonstrates that the pruned cells of the proposed backbone usually contains the Hadamard product introduced as a two-input operation, while the number of parameters involved in these pruned cells is on the same order with the baseline cells. INDEX TERMS DA-RNN, DARTS backbone, neural architecture search, stacked RNN, time series forecasting.