Figure 18 - uploaded by Mohammad Shahadat Hossain
Content may be subject to copyright.
Source publication
Investors in the stock market have always been in search of novel and unique techniques so that they can successfully predict stock price movement and make a big profit. However, investors continue to look for improved and new techniques to beat the market instead of old and traditional ones. Therefore, researchers are continuously working to build...
Similar publications
The development of the stock market in Indonesia is currently growing quite rapidly. This can be seen based on the number of investors who have increased every year. In 2011, sharia stocks were launched for the first time in Indonesia, and it can be seen that the price of the stock is not always stable or can experience increases or decreases. For...
![Block diagram of the state of art system: [13] [The blue borders show...](publication/356591638/figure/fig1/AS:1126118237192192@1645498918105/Block-diagram-of-the-state-of-art-system-13-The-blue-borders-show-the-good-features_Q320.jpg)
The prediction of stock price movement direction is significant in financial circles and academic. Stock price contains complex, incomplete, and fuzzy information which makes it an extremely difficult task to predict its development trend. Predicting and analysing financial data is a nonlinear, time-dependent problem. With rapid development in mach...
This study investigates the relationship between founding family ownership and firm performance in the Italian stock market. Making use of a precise definition of Founding family ownership factor, an empirical analysis on the stock monthly returns has been carried out, from an investor’s point of view facing an asset allocation problem. Portfolios...
The January effect anomaly that occurred in the Indonesian stock market was inconsistent and only occurred in a few years. However, its existence is sufficient to create a potential negative return risk in the non-January trading month. So this research purposes to identify the characteristics of the January effect anomaly that occurs and its effec...
Practice and empirical observations prove that achieving above-average returns on the stock market is possible. It is possible to achieve both higher and lower returns than those resulting from the fundamental value of the companies being valued. This condition is affected by anomalies that make the market ineffective. Numerous studies in behaviour...
Citations
... The same author also used decision trees in [36] to evaluate the effectiveness of 60 several inter-class and probabilistic distance-based methods in feature selection. As for the 61 stock market predictions, [4,11] used the OHLC prices and the Volume as inputs. In [9], 62 Dutta et. ...
... al. used 20 features to predict the price of bitcoin, including the hash rate, the 63 lag daily returns, the price volatility among others. In [10,22] As for analyzing the forecast window, [11] studied the prediction of 3 selected stocks 87 for 1 day, 3 days and 5 days, whereas [16] analyzed the impact of the train-test split. ...
In this study, we analyze the impact of input feature selection on the accuracy of short-term 1 price predictions of crude oil futures. We analyze the impact of using 39 different features compared 2 to using the Close price alone. The features we consider are the High, the Low, the Open and the 3 Close (OHLC) prices and the Volume, in addition to a list of technical indicators derived from them. 4 We use a model built from Bi-directional Gated Recurrent Units (BGRUs) for the analysis. In addition, 5 we quantitatively analyze the impact of varying the prediction window on the accuracy of prediction. 6 We find that adding more input features does not improve the accuracy of prediction compared 7 to using the Close price alone. On the contrary, we find evidence that using more input features 8 beyond the close price actually deteriorates the accuracy of prediction. We also find that the accuracy 9 of prediction is quickly lost once the prediction window exceeds about an hour into the future. 10 Therefore, it is recommended to confine the input features to the Close price alone, and limit the 11 prediction window to less than an hour for the purposes of day trade.
... Another comparative study was conducted by Dey et. al. in [13] where a simple 96 RNN model, an LSTM model and a GRU model were compared. The study compared 97 the performance of the 3 models on one-day, three-day and five-day time horizons of the 98 stock prices of 3 selected companies. ...
The prediction of the stock market and the prices of other commodities like crude oil, 1 constitutes a challenging task. Recently, the rapid progress in the field of Machine Leaning (ML), 2 led to an increased interest in applying ML techniques to market price predictions. In this study, 3 we conduct a comprehensive comparative analysis of the performance of 15 different ML models in 4 predicting the close prices of crude oil futures. These models include an Auto-Regressive Integrated 5 Moving Average (ARIMA) model, the Meta Prophet Library, a simple Recurrent Neural Network 6 (RNN), a Long-Short Term Memory (LSTM), a Gated Recurrent Unit (GRU), a Bi-directional LSTM 7 (BLSTM), a Bi-directional GRU (BGRU) and a number of hybrid models, including an LSTM-BGRU 8 and a BLSTM-BGRU. In addition, we evaluate the effectiveness of using Denoising Autoencoders 9 (DAE) in enhancing the performance of the networks by evaluating hybrid models with DAE layers, 10 including DAE-LSTM, DAE-BLSTM, DAE-GRU, DAE-BGRU, DAE-LSTM-GRU and DAE-BLSTM-11 BGRU. In our analysis, we focus on short-term predictions dedicated for day trading. We compare 12 the performance of these models using the Root Mean Square Error (RMSE), Mean Absolute Error 13 (MAE), Mean Absolute Error Percentage (MAPE) and the R 2. We find that the BGRU model yields the 14 best performance, with the GRU model not far behind. We also find that hybrid models containing 15 BGRUs or GRUs tend to perform better than other hybrid models. We find mixed evidence visa -vis 16 the effectiveness of DAEs in improving the performance of networks, where in some models the 17 performance improves, whereas for others the performance deteriorates. We find that the performance 18 of all models deteriorate when predicting sharp and sudden changes in prices. The key takeaway of 19 this study is that BGRUs, and to a lesser extent GRUs, seem to be the best route to follow in applying 20 AI/Ml to the task of market prediction. 21
... The same author also used decision trees in [36] to evaluate the effectiveness of 60 several inter-class and probabilistic distance-based methods in feature selection. As for the 61 stock market predictions, [4,11] used the OHLC prices and the Volume as inputs. In [9], 62 Dutta et. ...
... al. used 20 features to predict the price of bitcoin, including the hash rate, the 63 lag daily returns, the price volatility among others. In [10,22] As for analyzing the forecast window, [11] studied the prediction of 3 selected stocks 87 for 1 day, 3 days and 5 days, whereas [16] analyzed the impact of the train-test split. ...
for possible open access publication under the terms and conditions of the Creative Commons Attri-bution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/). Abstract: In this study, we analyze the impact of input feature selection on the accuracy of short-term 1 price predictions of crude oil futures. We analyze the impact of using 39 different features compared 2 to using the Close price alone. The features we consider are the High, the Low, the Open and the 3 Close (OHLC) prices and the Volume, in addition to a list of technical indicators derived from them. 4 We use a model built from Bi-directional Gated Recurrent Units (BGRUs) for the analysis. In addition, 5 we quantitatively analyze the impact of varying the prediction window on the accuracy of prediction. 6 We find that adding more input features does not improve the accuracy of prediction compared 7 to using the Close price alone. On the contrary, we find evidence that using more input features 8 beyond the close price actually deteriorates the accuracy of prediction. We also find that the accuracy 9 of prediction is quickly lost once the prediction window exceeds about an hour into the future. 10 Therefore, it is recommended to confine the input features to the Close price alone, and limit the 11 prediction window to less than an hour for the purposes of day trade. 12
... This characteristic makes RMSE useful for analyzing the error gap between the expected and predicted values [162]. Moreover, RMSE assigns a higher weight to larger errors compared to other measures, making it particularly suitable for domains where significant errors in accuracy are undesirable [161]. Incorporating the magnitude of error into RMSE is pertinent in stockprice prediction research, as larger errors in prediction can potentially lead to greater losses in buy or sell decisions. ...
Price prediction tools play a significant role in small investors’ behavior. As such, this study aims to propose a method to more effectively predict stock prices in North America. Chiefly, the study addresses crucial questions related to the relevance of news and tweets in stock-price prediction and highlights the potential value of considering such parameters in algorithmic trading strategies—particularly during times of market panic. To this end, we develop innovative multi-layer perceptron (MLP) and long short-term memory (LSTM) neural networks to investigate the influence of Twitter count (TC), and news count (NC) variables on stock-price prediction under both normal and market-panic conditions. To capture the impact of these variables, we integrate technical variables with TC and NC and evaluate the prediction accuracy across different model types. We use Bloomberg Twitter count and news publication count variables in North American stock-price prediction and integrate them into MLP and LSTM neural networks to evaluate their impact during the market pandemic. The results showcase improved prediction accuracy, promising significant benefits for traders and investors. This strategic integration reflects a nuanced understanding of the market sentiment derived from public opinion on platforms like Twitter.
... Purohit et al trained a RNN model on Google data-set and found it capable enough in predicting the near future [19]. Dey et al carried out a comparative analysis of RNN in stock prediction, concluding that it performed best when the pattern was relatively gentle without obvious abnormal fluctuations [20]. ...
As the global economy experiences unprecedented growth, the stock market is undergoing a period of expansion and prosperity. Driven by the allure of high risks and high returns, a significant number of individuals have committed themselves to stock trading. In the age of the Internet of Everything, the trajectory of the stock market is intertwined not only with the fortunes of individuals and businesses but also deeply influences the fate of nations and societies. However, stock market forecasting is quite a hard task owing to its volatile nature, where even seemingly unrelated factors can lead to dramatic changes. Therefore, the importance of constructing reliable stock prediction methods is self-evident. In recent years, scholars have mainly focused on machine learning and deep learning methods to complete this task. After literature review, this article summarizes and discussed several commonly used and effective methods, such as Linear Regression, Random Forest, Long Short-Term Memory (LSTM), etc. The advantages, disadvantages and applications of different methods are discussed, like the shortcoming of Linear Regression to handle non-linear models well and the relatively excellent ability of LSTM to predict long sequences. This article would be helpful for emerging scholars to have a grasp of the characteristics, advantages and scope of application of these methods, and facilitate the development of future research.
... The extracted features are given as input to the LSTM [40] for classifying the type of tumours. LSTM is chosen for its capacity for modelling the temporal dependencies, making it appropriate for sequential data i.e., time series. ...
... (36)-(40). Further, the SSIM, Jaccard, dice are numerically expressed in Eqs. (41)-(43). ...
Brain tumour is an abnormal growth of cells in the brain, and is a harmful and life-threatening disease worldwide. The rapid development of tumour cells increases the illness and death rates. Hence, the timely detection and classification of brain tumours is crucial for saving thousands of lives. However, the diagnosis of brain tumour is a challenging task because of different brain shapes, sizes and synaptic structures. In this research, the Linear Intensity Distribution Information (LIDI) and regularization parameter based Intuitionistic Fuzzy C-Means Algorithm (IFCM), namely LRIFCM is proposed for an effective segmentation of the brain portion that is affected by tumours. The different feature extraction approaches namely, LeNET, Gray-Level Co-occurrence matrix (GLCM) and Local Ternary Pattern (LTP) are used to extract appropriate features from the brain images. Further, the Long Short-Term Memory (LSTM) classifier is used to classify the types of tumour based on the extracted features. Three different datasets namely, BRATS 2017, BRATS 2018 and Figshare brain datasets are used to analyse the proposed LRIFCM-LSTM method. The LRIFCM-LSTM is evaluated using both the segmentation and classification results. The segmentation using LRIFCM is assessed based on the parameters of SSIM, Jaccard, dice, accuracy and sensitivity, whereas the classification using LSTM is assessed based on accuracy, specificity, sensitivity, precision and F1-score. The existing researches: Hybrid-DANet, TECNN and VAE-GAN are used for comparison with the LRIFCM-LSTM method. The classification accuracy of LRIFCM-LSTM for BRATS 2018 dataset is 98.73 which is higher when compared to the TECNN.
... However, with the advent of big data and advances in learning methods, deep learning technology is also being used in stock market forecasting, and its learn-ing ability is higher than that of traditional machine learning [26]. Among all deep learning technologies, Recurrent Neural Networks (RNNs), including the simple RNN, Long Short-Term Memory (LSTM), and Gated Recurrent Unit (GRU), are most commonly employed for financial time series forecasting due to their ability to capture time dependencies and sequential patterns in data [27]. Demonstrating the effectiveness of these techniques, Satria, D. [28] conducted a comparative study involving the simple RNN, LSTM, and GRU to predict stock prices in the banking sector in Indonesia. ...
... Fig. 2 shows the architecture of the simple RNN in detail, and the hidden states of the simple RNN are given by eqs. (1) and (2) [27]. ...
This study presents a new approach that combines Data Envelopment Analysis (DEA) for stock selection based on fundamental metrics with Recurrent Neural Networks (RNNs), including simple RNN and Long Short-Term Memory (LSTM) networks, to predict future stock trends in the Thai stock market. DEA can identify the most qualified stocks based on fundamental metrics, while RNNs analyze the historical prices of these selected stocks to forecast their future trends. This innovative DEA-RNNs approach aims to improve the reliability of the investment decision-making process. Using the DEA methodology, our study evaluated thirty-seven stocks based on their financial metrics from the first quarter of 2017 to the third quarter of 2022. Following the analysis, DEA identified ten stocks as the most promising. Subsequent predictive modeling using RNNs showed that only seven out of these ten initially identified stocks were projected to experience an upward trend in the subsequent quarter. To assess the predictive models' performance, we utilized metrics such as mean square error (MSE) and mean absolute error (MAE), consistently demonstrating the LSTM model's superior performance over the simple RNN model.
... These authors were not alone in their general findings, as others found in a comparative analysis of Recurrent Neural Networks, that LSTM and GRU outperformed the simple RNN. Specifically, GRU performed better for stocks with high fluctuations, while LSTM excelled when the price patterns were moderately variable (Dey et al., 2021). ...
This study delves into the efficacy of various machine learning and statistical models that have captured the attention of financial analysts. Two of them, Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU) are variations of Recurrent Neural Networks while the Autoregressive Integrated Moving Average (ARIMA) is a statistical model. These models will be used to forecast stock market data across different economic sectors. In the dynamic landscape of financial markets, accurate forecasting is crucial. This research paper contributes to quantitative finance by conducting a comprehensive comparative analysis of these models on historical stock market data from three sectors: extraction, manufacturing, and service (which are considered the primary, secondary, and tertiary sectors of the economy respectively). The models' performances are evaluated using mean squared error (MSE) on six selected stocks representing these sectors. Results reveal the power of recurrent neural networks in capturing intricate patterns. Moreover, the results will explore whether or not the efficacy of each model is impacted by the sector of the economy that it is forecasting data for.
... The companies sell stocks on a stock market to extend their businesses and the investors or buyers can buy the companies stocks and sell those stocks to other investors at a predetermined rate. The investors can get the profit by selling those stocks at any time with an additional cost [2]. Stock price fluxes are connected with several variables, including market expectation, macroeconomic circumstances, and trust in the concern's managing and maneuvers. ...
The stock market has considered the active research fields today, and forecasting its behaviour is an enormous necessity. Predicting the stock market is complex, necessitating a thorough examination of data patterns. Correct forecasting outcomes can provide significant insight to investors, lowering investment risk. This paper proposes a novel Weight and Bias Tuned Long Short-term Memory (WBTLSTM) with an efficient feature extraction model, Bilateral ReLU-based Two-Dimensional Convolutional Neural Network (BR2DCNN), for stock price prediction (SPP). First, the stock data was collected from the publicly available dataset. Then the missing values imputation and data normalization is performed on the collected dataset. The preprocessed dataset extracts the most relevant features using BR2DCNN. Finally, the future stock events are predicted using the WBTLSTM. The weights and biases are tuned with the help of the Enhanced Butterfly Optimization Algorithm (EBOA). Experimental findings prove that the proposed one achieves superior outcomes compared to the conventional methods regarding some performance metrics.
... The need to use LSTMs instead of RNNs stems from the "vanishing" gradient problem, which is the possibility that the gradient will vanish or explode during the back-propagation operation, which is the true pillar of artificial neural network training [28]. LSTM networks can be used for either classifying temporal sequences or predicting successive temporal instants, depending on the output layer chosen. ...
Having become aware of how limited all natural resources are, the water leakage problem in piping systems has become a stimulating topic. This problem increased over the last few years even though innovative tools and techniques appeared in the literature and in the consumer market. Identifying water leaks at the nearest point, the household level, is still an unsolved problem because most water meters are mechanical and, therefore, cannot detect leaks. While the issue is not important for water service providers since consumption is charged to the user, the resolution is crucial due to the increasingly relevant concern of saving natural resources. The detection of small but continuous leaks of drinking water in domestic systems is addressed in this work. Machine learning approaches enabled image processing techniques also in uncontrolled environments, overcoming the classical methods but introducing new challenges like power consumption. Using a combination of Convolutional Neural Networks (CNNs) and Long Short–Term Memory Networks (LSTM) within an adaptive under–sampling strategy, it is possible to process the images captured from the mechanical water meter dial and identify the Period With Null Consumption (PWNC) or the consumption class. The presented solution can classify the water flow into four different classes, and, in the case of absence or small flow, its function becomes to detect leakages. Analyzing images from a mechanical water meter quadrant, it has been possible to identify PWNC and detect small water leakages in the domestic environment under common consumer flow profiles. In addition to the confusion matrices, the synthetic parameters of Sørensen–Dice coefficient (DSC) and Jaccard Index have been utilized and presented to quantify the performance of the proposed DNN. The conducted experiments on static and dynamic water flow demonstrated the applicability of this approach and the possibility of an increase in PWNC identification thanks to the adaptative increase in the sampling time. Moreover, the reduction of sampling time allows for the reduction of computational load and power consumption in embedded scenarios where limited energy is available.
