Stock Price Forecasting Research Articles

Predicting Stock Prices Using Deep Learning Algorithms: A Case of Food-Processing Industry

Prediction of stock prices has been a focal point of the financial body of knowledge for decades now. The complexity of stock price prediction involves various factors, including market trends, economic indicators, company performance, news sentiment, and more. Accordingly, stock prices are said to follow the ‘random walk hypothesis’. This systemic factor should be coupled with the limited human cognitive abilities to envisage the dynamics of financial markets. Novel machine learning algorithms have been advocated as potentially supreme replacement for the ‘human-centric’ stock prediction approaches. Hitherto, a myriad of machine learning algorithms has been effectively used for this purpose – ARIMA (Auto-Regressive Integrated Moving Average), XGBoost (Random Forest and Gradient Boosting Algorithms), CNN (Convolutional Neural Networks) or LSTM (Long Short-Term Memory). The aim of this paper is to test the predictive capacity of LSTM on a sample of large global food industry companies. The prices of shares of five companies were observed, namely: PEP (PepsiCo), TSN (Tyson Foods), NSRGY (Nestle), JBSAY (JBS S.A.), KHC (The Kraft Heinz Company), in the period from 01.01.2015. until 1.11.2022. Based on the data from this time range, a stock price forecast for Nov 2nd, 2022, was made. The results indicate very precise prediction since the difference between predicted and real stock price is insignificant. Keywords: stock price, machine learning, long-short term memory, food-processing industry

The volatility mechanism and intelligent fusion forecast of new energy stock prices

AbstractThe new energy industry is strongly supported by the state, and accurate forecasting of stock price can lead to better understanding of its development. However, factors such as cost and ease of use of new energy, as well as economic situation and policy environment, have led to continuous changes in its stock price and increased stock price volatility. By calculating the Lyapunov index and observing the Poincaré surface of the section, we find that the sample of the China Securities Index Green Power 50 Index has chaotic characteristics, and the data indicate strong volatility and uncertainty. This study proposes a new method of stock price index prediction, namely, EWT-S-ALOSVR. Empirical wavelet decomposition extracts features from multiple factors affecting stock prices to form multiple sub-columns with features, significantly reducing the complexity of the stock price series. Support vector regression is well suited for dealing with nonlinear stock price series, and the support vector machine model parameters are selected using random wandering and picking elites via Ant Lion Optimization, making stock price prediction more accurate.

Stock Price Prediction Using Candlestick Patterns and Sparrow Search Algorithm

Accurately forecasting the trajectory of stock prices holds crucial significance for investors in mitigating investment risks and making informed decisions. Candlestick charts visually depict price information and the trends in stocks, harboring valuable insights for predicting stock price movements. Therefore, the challenge lies in efficiently harnessing candlestick patterns to forecast stock prices. Furthermore, the selection of hyperparameters in network models has a profound impact on the forecasting outcomes. Building upon this foundation, we propose a stock price prediction model SSA-CPBiGRU that integrates candlestick patterns and a sparrow search algorithm (SSA). The incorporation of candlestick patterns endows the input data with structural characteristics and time series relationships. Moreover, the hyperparameters of the CPBiGRU model are optimized using an SSA. Subsequently, the optimized hyperparameters are employed within the network model to conduct predictions. We selected six stocks from different industries in the Chinese stock market for experimentation. The experimental results demonstrate that the model proposed in this paper can effectively enhance the prediction accuracy and has universal applicability. In comparison to the LSTM model, the proposed model produces an average of 31.13%, 24.92%, and 30.42% less test loss in terms of MAPE, RMSE and MAE, respectively. Moreover, it achieves an average improvement of 2.05% in R2.

The Influence of Social Media on Stock Market: A Transformer-Based Stock Price Forecasting with External Factors

This paper introduces an innovative approach to forecasting stock prices. Forecasting stock prices is crucial in assisting investors in making informed decisions. Our research presents a unique method that utilizes transformer-based machine learning approach for stock price forecasting. This method exploits self-attention mechanisms to grasp intricate patterns and dynamics within historical stock price data. To bolster our model performance, we integrate investors’ sentiment collected from social media by using sentiment analysis with the help of natural language processing. Utilizing the variation caused by investors’ sentiment over time, as well as external macroeconomic factors, our proposed model outperforms benchmark models. Through extensive comparisons with various benchmark machine-learning algorithms, results produced by our proposed method are favorably comparable to those produced by conventional approaches. Across multiple machine learning models, our preferred model demonstrates superior performance, achieving an RMSE value of 0.96 compared to the RMSE value of 1.58 obtained from LSTM model.

A novel hierarchical feature selection with local shuffling and models reweighting for stock price forecasting

Stock price forecasting is a challenging task due to the complexity of financial markets and the high volatility of stocks. Because of the strong nonlinear representation ability of neural network models, such as long-short memory network (LSTM) and deep learning, they are used extensively in recent years for stock price forecasting. However, due to the volatility of financial data, neural network models often suffer from overfitting or instability problems. In addition, quantitative trading with feature mining tools can generate a growing number of features for financial data. Therefore, selecting effective features for financial data is an urgent problem. To address these problems, we propose a novel hierarchical feature selection with local shuffling (HFSLS) and models reweighting (MR) based on LSTM, named HFSLSMR-LSTM, for stock price forecasting. Specifically, for each layer, local shuffling perturbs each feature to re-predict, and its predicted value is compared with the true value to calculate the feature importance, and the important features are selected and returned to the next layer. Besides, a proximity reweighting scheme is presented to adjust the weight for each layer model that learns from hierarchical features. The HFSLSMR-LSTM model is still effective for financial data with multiple features and frequent fluctuations. Experimental results on stock index dataset and Dow Jones Industrial Average dataset demonstrate that the HFSLSMR-LSTM outperforms Informer, DoubleEnsemble, LSTM, GRU, BI-GRU and BI-LSTM on the metrics MSE, RMSE, MAE, MAPE and R2.

An interval constraint-based trading strategy with social sentiment for the stock market

Developing effective strategies to earn excess returns in the stock market is a cutting-edge topic in the field of economics. At the same time, stock price forecasting that supports trading strategies is considered one of the most challenging tasks. Therefore, this study analyzes and extracts news media data, expert comments, social opinion data, and pandemic text data using natural language processing, and then combines the data with a deep learning model to forecast future stock price patterns based on historical stock prices. An interval constraint-based trading strategy is constructed. Using data from several typical stocks in the Chinese stock market during the COVID-19 period, the empirical studies and trading simulations show, first, that the sentiment composite index and the deep learning model can improve the accuracy of stock price forecasting. Second, the interval constraint-based trading strategy based on the proposed approach can effectively enhance returns and thus, can assist investors in decision-making.

The Non-Seasonal Holt-Winters Method for Forecasting Stock Price Returns of Companies Affected by BDS Action

The non-seasonal Holt-Winters method is one of the methods of smoothing theory. This method can be implemented on time series data that does not have a seasonal component. In this study, this method is used to forecast the stock price returns of companies affected by the Boycott, Divestment, and Sanctions (BDS) action. Forecasting gets very good results that can be seen from the MAPE value of modeling the six stocks affiliated with Israel that continue to carry out Zionism against Palestine is not more than 10%. This method can also accommodate the limitations of existing data while still obtaining good forecasting results. In addition, the use of several transformations of stock price returns in this case is very useful in modeling to obtain appropriate error assumptions. The forecasting results of the model formed as a whole follow the trend in the stock price of each company. To produce good forecasting results using this method, it is recommended to do forecasting in the short term. The forecasting results show that of the six company stocks, almost all of them experienced a decrease in stock price returns. Only one stock of PT Map Boga Adiperkasa Tbk has increased. This also illustrates that the BDS action influences on these companies.

Beyond traditional networks: Stock predictions via enhanced backpropagation

Predicting stock prices has long been a challenge due to the inherent volatility and intricacies of stock markets, and addressing these forecasting challenges has broad implications for the financial sector and those involved in market activities. This research holds pivotal importance as it equips traders, investors, and financial institutions with enhanced tools, facilitating better decision-making and optimized strategies. Beyond its academic significance, an accurate model becomes an indispensable asset in the real-world navigation of stock markets, especially in a realm where slight prediction discrepancies can result in substantial financial impacts. This study endeavours to introduce and validate an enhanced Backpropagation neural network with the core objective of elevating stock price prediction accuracy and reliability to a benchmark level. Adopting a meticulously crafted Backpropagation neural network, designed specifically for improved accuracy in stock price forecasting, we employed rigorous evaluation methods, measuring the model's performance against key metrics MAE and MSE. Additionally, visual representations were given to provide a more intuitive understanding of the model's prowess. The results were clear: the enhanced model demonstrably excelled in prediction precision, as evidenced by the favourable MAE and MSE outcomes. Visual narratives further accentuated its adeptness at tracing the complex oscillations inherent to stock market behaviours, underscoring both its academic contribution and potential as a transformative tool in the practical landscape of stock forecasting.

Stock price forecasting in Indonesia stock exchange using deep learning: a comparative study

In 2022, the Indonesia stock exchange (IDX) listed 825 companies, making it challenging to identify low-risk companies. Stock price forecasting and price movement prediction are vital issues in financial works. Deep learning has previously been implemented for stock market analysis, with promising results. Because of the differences in architecture and stock issuers in each study report, a consensus on the best stock price forecasting model has yet to be reached. We present a methodology for comparing the performance of convolutional neural networks (CNN), gated recurrent units (GRU), long short-term memory (LSTM), and graph convolutional networks (GCN) layers. The four layers types combination yields 11 architectures with two layers stacked maximum, and the architectures are performance compared in stock price predicting. The dataset consists of open, highest, lowest, closed price, and volume transactions and has 2,588,451 rows from 727 companies in IDX. The best performance architecture was chosen by a vote based on the coefficient of determination (R2), mean squared error (MSE), root mean square error (RMSE), mean absolute percent error (MAPE), and f1-score. TFGRU is the best architecture, producing the finest results on 315 companies with an average score of RMSE is 553.327, MAPE is 0.858, and f1-score is 0.456.

Stock Price Prediction Using Sentiment Analysis

Abstract: This research discusses the ever-changing nature of the stock market by using models such as LSTM and ARIMA to forecast stock prices. It recognizes the volatility and unpredictability of financial markets recognizing the need for robust tools that can accurately predict. Sentiment analysis is used for enhancing the accuracy of stock price prediction, drawing information from various sources like news resources and social media platforms that reflect public sentiments and opinions. We aim to analyze a thorough aspect of factors influencing stock prices by combining these sentiments with the models. The LSTM model is used to study long-term dependencies in stock price trends and ARIMA provides information into the time series components. Combining sentiment analysis helps us to scan the emotional tone around a particular stock and contributes valuable data to the prediction process. The integration between the machine learning models and sentiment analysis offers an extensive approach to predicting stock prices considering both past trends and current public sentiment. This study adds to the work in progress to develop more accurate and adaptive tools for exploring the difficulties of the stock market.

Harnessing Cognitively Inspired Predictive Models to Improve Investment Decision-Making

In the last years, researchers and practitioners have focused on defining portfolio optimization approaches. This task aims to identify a suitable distribution of assets for maximizing profits and minimizing risks, also offering protection against unexpected market behaviors. Nevertheless, the state-of-the-art approaches encounter significant limitations due to the complex nature of the task: (1) forecasting of non-stationary, non-linearity and volatile stock price; (2) budget allocation over different stocks satisfying multi-objective objective function; (3) risk costs can significantly affect the effectiveness of the designed approaches. In this paper, we propose a cognitively inspired framework for portfolio optimization by integrating deep learning-based stock forecasting for maximizing the revenue and portfolio diversification and Shape Ratio for minimizing the risk. Furthermore, the cognitively inspired forecasting module relies on the LSTM-based approach which combines historical financial data and technical indicators. Hence, this approach addresses the portfolio optimization task with the aim of designing more and more cognitive agents that perform autonomous actions for supporting decision-making. To make these agents cognitive, we further integrate stock forecasting into the portfolio optimization model, also investigating the main factors affecting both stock forecasting and portfolio optimization tasks. The proposed framework has been evaluated in two stages on a real-world dataset, composed of four years of information about stocks from six different areas. Firstly, we compare the proposed forecasting models based on LSTM and GRU, pointing out that the former achieves higher effectiveness results although the latter has a shorter training time. Finally, the proposed framework has been compared with different baselines, obtaining a net difference of $168 at the maximum. Finally, we compare the proposed approach w.r.t. several baselines in terms of total revenue, also providing an ablation analysis to investigate how stock prediction might support investors in dealing with portfolio optimization task.

Stock Market Prediction using Machine Learning

Stock market prediction has been a subject of significant interest and research for both financial analysts and machine learning practitioners. This abstract presents a concise overview of the key aspects and approaches in the realm of stock market prediction. The unpredictable and dynamic nature of financial markets poses a challenge for accurate forecasting. However, advancements in machine learning techniques, availability of large-scale financial data, and computational power have led computational to the development of sophisticated prediction models. In this endeavour, we investigate the application of various machine learning algorithms, including regression, time series models and support vector machine, to forecast stock prices. The research focuses on data preprocessing, feature engineering, and model evaluation to enhance prediction accuracy. Using a diverse dataset Evaluation metrics such as Mean Squared Error (MSE), Root Mean Squared Error (RMSE), and Mean Absolute Percentage Error (MAPE) are utilized to measure model performance. While acknowledging the inherent uncertainty of financial markets, this research contributes to the broader dialogue on data-driven decision-making in investment and finance. The outcomes of this study offer insights into the strengths and limitations of machine learning techniques in stock price prediction.

Willingness to Use Algorithms Varies with Social Information on Weak vs. Strong Adoption: An Experimental Study on Algorithm Aversion

The process of decision-making is increasingly supported by algorithms in a wide variety of contexts. However, the phenomenon of algorithm aversion conflicts with the development of the technological potential that algorithms bring with them. Economic agents tend to base their decisions on those of other economic agents. Therefore, this experimental approach examines the willingness to use an algorithm when making stock price forecasts when information about the prior adoption of an algorithm is provided. It is found that decision makers are more likely to use an algorithm if the majority of preceding economic agents have also used it. Willingness to use an algorithm varies with social information about prior weak or strong adoption. In addition, the affinity for technological interaction of the economic agents shows an effect on decision behavior.

Explore the Impact of Initial Data Coherence on ARIMA Model Prediction Based on Python

As the most intuitive economic indicator reflecting the economic development level of a company, an industry, or even a country, the fluctuation of stock prices will attract great attention. In order to avoid the loss of investors to a certain extent, stock price prediction is widely used in the stock market to prevent risks to a certain extent, and the ARIMA model is a time series prediction model that has been widely used in recent years. Therefore, the application of the ARIMA model to stock price forecasting is the main research object of this paper. This paper speculated at the beginning of the study that there may be a close relationship between the degree of fit of the ARIMA model and the consistency of the prediction results with the original data used. However, the existing literature is basically blank in this aspect of research and analysis. Therefore, this paper chooses to conduct comparative experiments on a set of continuous data and a set of discontinuous data to analyze the influence of the consistency of the original data under application conditions on the ARIMA model fitting and prediction results, hoping to fill the gaps in the existing literature through this research. The experimental results show that in some cases of original data with obvious discontinuous data missing, its smoothness is extremely poor, resulting in a poor fitting effect of the ARIMA model, and it can not be applied to the prediction of such data.

Comparison of the Accuracy of Brown's and Holt's Double Exponential Smoothing in LQ45 Stock Price Forecasting

As of May 2022, 787 stocks are listed on the Indonesia Stock Exchange (IDX), and the number of stock indices in Indonesia to date is 38. One interesting and important stock index is the LQ45 index. Because this index is a very important reference index for investors, this research data focuses on stocks in the LQ45 index. There are two essential things in the forecasting process: the data and the right forecasting method. Two forecasting methods that can be used are Brown and Holt's Double Exponential Smoothing (DES). This study examines two methods with the lowest accuracy error in forecasting the LQ45 stock price data. Mean Absolute Percentage Error (MAPE) is used to measure the accuracy of the error. The analysis methods used to compare the MAPE of the two methods are the F test for variance similarity, Boxplot, t-test to test paired means with different cases of variance, and Wilcoxon signed rank test to test paired means nonparametric statistics. The result is that the MAPE average with Holt's DES method is smaller than the average MAPE with Brown's DES method. This is supported by the t-test for paired means with different cases of variance and also supported by the Wilcoxon signed exact rank test. Meanwhile, the MAPE standard deviation with Holt's DES method is smaller than the MAPE standard deviation with Brown's DES method. This is supported by the F test to test the variance similarity and is visually supported by a Boxplot diagram. From this study, LQ45 stocks with the smallest MAPE value accuracy are ICBP stocks. In general, based on the MAPE value, Holt's DES method is better than Brown's DES method in predicting the prices of stocks in the LQ45 index.

Algorithm Optimizer in GA-LSTM for Stock Price Forecasting

Fluctuating stock prices make it difficult for investors to see investment opportunities. One tool that can help investors overcome this is represented by forecasting techniques. Long Short-Term Memory (LSTM) is one of deep learning methods used in forecasting time series. The training and success of deep learning is strongly influenced by the selection of hyperparameters. This research uses a hybrid method between the Genetic Algorithm (GA) and LSTM to find a suitable model for predicting stock prices. GA is used in optimizing the architecture such as the number of epochs, window size, and the number of LSTM units in the hidden layer. Tuning optimizer is also carried out using several optimizers to achieve the best value. From method that has been applied, it shows that the method has a good level of accuracy with MAPE values below 10% in every optimizer used. The error rate generated is quite low, in case-1 with a minimum RMSE value of 93.03 and 94.40, & in case-2 with an RMSE value of 104.99 and 150.06 during training and testing. A fairly stable and small value is generated by setting it using the Adam Optimizer.

A bibliometric literature review of stock price forecasting: From statistical model to deep learning approach.

We introduce a comprehensive analysis of several approaches used in stock price forecasting, including statistical, machine learning, and deep learning models. The advantages and limitations of these models are discussed to provide an insight into stock price forecasting. Traditional statistical methods, such as the autoregressive integrated moving average and its variants, are recognized for their efficiency, but they also have some limitations in addressing non-linear problems and providing long-term forecasts. Machine learning approaches, including algorithms such as artificial neural networks and random forests, are praised for their ability to grasp non-linear information without depending on stochastic data or economic theory. Moreover, deep learning approaches, such as convolutional neural networks and recurrent neural networks, can deal with complex patterns in stock prices. Additionally, this study further investigates hybrid models, combining various approaches to explore their strengths and counterbalance individual weaknesses, thereby enhancing predictive accuracy. By presenting a detailed review of various studies and methods, this study illuminates the direction of stock price forecasting and highlights potential approaches for further studies refining the stock price forecasting models.

ARIMA with Attention-based CNN-LSTM and XGBoost hybrid model for stock prediction in the US stock market

Absteact: Stock price forecasting is considered one of the most difficult tasks in financial forecasting. Combining ARIMA with neural networks helps to enhance the model’s predictive capabilities when dealing with complex, nonlinear time series data. Attention-based CNN-LSTM and XGBoost hybrid model achieves the accuracy of stock prediction results. However, the predictive effect of this hybrid model has only been confirmed by Chinese stock market data. Therefore, this paper proposes to use ARIMA with Attention-based CNN-LSTM and XGBoost hybrid model to predict the stock price of five different types of companies in the US stock market. Experimental results show that this hybrid model performs well in the training phase and has small prediction errors. However, during the testing phase, the model performs better in predicting stocks with larger price fluctuations, while the prediction error for stocks with more stable prices increases significantly. This shows that this hybrid model has good predictive ability when dealing with stocks with large market fluctuations, but it still needs further optimization for stable stocks. This study further demonstrates the potential of this hybrid model in stock price prediction, especially during market fluctuations, as it can be used as an effective tool to avoid risks and obtain returns.

Forecasting stock prices using a novel filtering-combination technique: Application to the Pakistan stock exchange

<abstract><p>Traders and investors find predicting stock market values an intriguing subject to study in stock exchange markets. Accurate projections lead to high financial revenues and protect investors from market risks. This research proposes a unique filtering-combination approach to increase forecast accuracy. The first step is to filter the original series of stock market prices into two new series, consisting of a nonlinear trend series in the long run and a stochastic component of a series, using the Hodrick-Prescott filter. Next, all possible filtered combination models are considered to get the forecasts of each filtered series with linear and nonlinear time series forecasting models. Then, the forecast results of each filtered series are combined to extract the final forecasts. The proposed filtering-combination technique is applied to Pakistan's daily stock market price index data from January 2, 2013 to February 17, 2023. To assess the proposed forecasting methodology's performance in terms of model consistency, efficiency and accuracy, we analyze models in different data set ratios and calculate four mean errors, correlation coefficients and directional mean accuracy. Last, the authors recommend testing the proposed filtering-combination approach for additional complicated financial time series data in the future to achieve highly accurate, efficient and consistent forecasts.</p></abstract>

The comparison stateless and stateful LSTM architectures for short-term stock price forecasting

Deep learning techniques are making significant contributions to the rapid advancements in forecasting. A standout algorithm known for its ability to produce accurate forecasts by recognizing temporal autocorrelation within the data is the Long Short-Term Memory (LSTM) algorithm, a component of Recurrent Neural Networks (RNN). The LSTM method employs both stateless and stateful architecture approaches, providing versatility in its application. This research aims to compare stateful and stateless algorithms in LSTM models, focusing on forecasting stock prices, such as those of Apple Inc. This comparative analysis is crucial, taking into account various characteristics of time series data, including the benefits and drawbacks of temporal autocorrelation. The comparison results reveal that, despite the stateful algorithm requiring more computational time, it achieves greater accuracy than the stateless approach. The forecast indicates a potential upward trend in share prices for the period of January to December 2024, according to the projected outlook for Apple's stock value. However, it is essential to exercise prudence in interpreting these results, considering that share price fluctuations are influenced by a significant number of variables.