Fig 4 - available via license: Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International
Content may be subject to copyright.
Source publication
In this paper, the daily returns of the S&P 500 stock market index are predicted using a variety of different machine learning methods. We propose a new multinomial classification approach to forecasting stock returns. The multinomial approach can isolate the noisy fluctuation around zero return and allows us to focus on predicting the more informa...
Contexts in source publication
Context 1
... support vector machines originally presented by Vapnik 37 can be used for both classification and regression problems. The basic idea and the terminology of support vector machines can be illustrated using a two-class classification problem with a linear decision boundary. The left-hand side of Fig. 4 illustrates the case with perfect separability. The right panel in Fig. 4 shows the nonseparable case, where some data points are misclassified by the linear decision boundary. The solid blue line in Fig. 4 is the decision boundary separating the two ...
Context 2
... originally presented by Vapnik 37 can be used for both classification and regression problems. The basic idea and the terminology of support vector machines can be illustrated using a two-class classification problem with a linear decision boundary. The left-hand side of Fig. 4 illustrates the case with perfect separability. The right panel in Fig. 4 shows the nonseparable case, where some data points are misclassified by the linear decision boundary. The solid blue line in Fig. 4 is the decision boundary separating the two ...
Context 3
... of support vector machines can be illustrated using a two-class classification problem with a linear decision boundary. The left-hand side of Fig. 4 illustrates the case with perfect separability. The right panel in Fig. 4 shows the nonseparable case, where some data points are misclassified by the linear decision boundary. The solid blue line in Fig. 4 is the decision boundary separating the two ...
Context 4
... scheme between the K(K À 1)/2 binary classifiers constructed for each class pair. For more information on the multiclass classification with support vector machines see eg., Hsu and Lin. 38 The goal with support vector machines is to find the decision boundary with maximum area on both sides of the boundary also known as the margin M ¼ 2 kbk . In Fig. 4 the dashed lines illustrate the margin and the points located at the dashed lines are known as support vectors. Hastie et al. 34 show that instead of maximizing the margin the optimization problem can be written in terms of minimizing ...
Similar publications
The dynamics and complexity of the stock market make it difficult for investors to form a stable and high-yield trading strategy. To meet such challenges, a novel stock trading strategy is proposed to help investors do better decisions in this paper. This strategy involves two main steps: turning point classification and reinforcement learning modu...
Rule-based portfolio construction strategies are rising as investment demand grows, and smart beta strategies are becoming a trend among institutional investors. Smart beta strategies have high transparency, low management costs, and better long-term performance, but are at the risk of severe short-term declines due to a lack of Risk Control tools....
Forecasts of regional electricity net-demand, consumption minus embedded generation, are an essential input for reliable and economic power system operation, and energy trading. While such forecasts are typically performed region by region, operations such as managing power flows require spatially coherent joint forecasts, which account for cross-r...
Citations
... A comprehensive survey of literatures elucidates that decision tree algorithms, exemplified by the Gradient Boosting Decision Tree (GBDT) technique, are also efficacious in yielding superior predictive outcomes in the domain of stock index forecasting. Undertaken by Nevasalmi (2020), an array of diverse machine learning models are applied to formulate predictions for the S&P 500 index. The empirical findings of this unequivocally demonstrate the superior predictive prowess of the Gradient Boosting Decision Tree (GBDT) model in comparison with conventional methodologies, such as Artificial Neural Networks (ANN) and Support Vector Machines (SVM). ...
In recent years, the challenging global economy has induced a surge in the investment risk. Precious metal index has gained popularity among investors because of its ability to diversify investments and hedge risk. To facilitate optimal decision-making in the face of market volatility, financial institutions, investors and related enterprises need an efficient and reliable method to predict the price fluctuations of precious metals. Based on these, the primary objective of this study is to develop an ensemble model which could predict precious metal index better. In this study, K-means method is employed for label segmentation, categorizing the price percentage changes of precious metal index into five distinct labels, and four machine learning models are selected for predictive performance comparison: XGBoost, LightGBM, CatBoost, and Random Forests. Then three models with better predictive performance are selected for ensemble. The ensemble model shows improved prediction ability compared to these individual ones and the empirical evidence is provided to demonstrate its robustness. In addition, this study provides an interpretable analysis of the characteristics with the SHAP method and proposes some investment recommendations. These results could provide ideas to the participants in the precious metal market.
... In the "Efficient Markets Hypothesis" developed by Fama's (1970) it was stated that the price of a stock fully reflects available information, therefore, the price movements of stocks are independent of their past prices and predicting prices in advance is impossible [3]. The validity of the Efficient Markets Hypothesis has been debated in many studies suggesting the inefficiency of markets [4]. As an alternative to the Efficient Markets Hypothesis, "The Adaptive Market Hypothesis" was developed. ...
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 %.
... However, many empirical studies have been conducted on the validity of this hypothesis. While the view that markets are efficient was supported in studies conducted in the 1960s and 1970s, studies conducted since the 1980s have supported the view that prices in markets are not independent of past information and therefore, markets are not efficient (Nevasalmi, 2020). In addition to these discussions, two different analysis methods, fundamental and technical analysis, are used to predict stock prices in stock markets. ...
The aim of this study is to compare the performances of the methods by predicting the movement directions of the stock market indices of developing countries with machine learning classification methods and to determine the best estimation method. For this purpose, the main stock market indices of E-7 countries, known as developing countries; Application was made on SSE (China), BOVESPA (Brazil), RTS (Russia), NIFTY 50 (India), IDX (Indonesia), IPC (Mexico) and BIST 100 (Turkey). k-nearest neighbors, decision trees, random forests, naive Bayes, support vector machines, artificial neural networks and logistic regression.methods, which are among the machine learning classification methods, were used to predict the movement directions of Stock Indices. In the analyses, the daily data for the period 01.01.2012-31.12.2021 and the technical indicators calculated based on the aforementioned data set were used as the data set. According to the results of the analysis, it has been determined that the best method for estimating the movement directions of the stock market indices of developing countries is logistic regression. Along with the logistic regression method, artificial neural networks and support vector machine methods were also determined to predict the direction of movement of all indices with an accuracy of over 70%. It has also been determined that the best method, logistic regression, is not valid for all indices, in other words, it is not the method with the highest accuracy performance in all indices.
... A comprehensive survey of literatures elucidates that decision tree algorithms, exemplified by the Gradient Boosting Decision Tree (GBDT) technique, are also efficacious in yielding superior predictive outcomes in the domain of stock index forecasting. Undertaken by Nevasalmi (2020), an array of diverse machine learning models are applied to formulate predictions for the S&P 500 index. The empirical findings of this unequivocally demonstrate the superior predictive prowess of the Gradient Boosting Decision Tree (GBDT) model in comparison with conventional methodologies, such as Artificial Neural Networks (ANN) and Support Vector Machines (SVM). ...
... When it comes to financial ratios, there are a lot of conflicting outcomes, forcing consumers to compare ratios and decide. Nevasalmi (2020) suggests that the ability to predict stock returns from its financial statement accurately is critical for building trading by hedging against probable market risks, speculators might benefit from stock index trading. According to Rjumohan (2019), forecasting a company's stock returns is very important because its high value most likely indicates the profit of its shareholders. ...
The rapid development and market rivalry eventually posed massive challenges in determining the future development trend of organizations to become market winners. Several types of research were conducted exploring the factors affecting investment in telecommunication companies. One has found significant relevance in financial management, which consists of financing, investment, and dividend decisions in private sector mobile telecom businesses. This study aimed to examine the impact of financial performance on the stock price of telecommunication companies listed on the IDX (Indonesia Stock Exchange) in the period of 2014-2020. The financial fundamentals used are current ratio (liquidity), return on asset (profitability), total asset turnover ratio (activity), and debt to asset ratio (solvency). This research applies qualitative data method. The population in this study is nine telecommunication companies listed on the IDX (Indonesia Stock Exchange) for seven-year period, and the sample selection technique used was purposive sampling. It resulted in 63 samples in total being obtained in this study. Moreover, the data type of this research uses secondary data, which is the annual financial report. The data analysis method in this study is descriptive analysis, classical assumption test, and multiple linear regression analysis processed using e-views 12. This research result is based on the fixed-effect method run since it is more suitable for the model of this research. The result shows that the financial fundamentals variables mutually have a significant impact on the stock performance. Yet, not all variables are significant individually.
... Analysis of mix of techniques very complex. The data taken is generally from 6 to 28 years[61,63]. The strained gradient-based technique combined with the Mamdani-type fuzzy system can be used in future research ...
This paper systematically reviews the literature related to stock price prediction systems. The reviewers collected 6222 research works from 12 databases. The reviewers reviewed the full-text of 10 studies in preliminary search and 70 studies selected based on PRISMA. This paper uses the PRISMA-based Python framework systematic-reviewpy to conduct this systematic review and browser-automationpy to automate downloading of full-texts. The programming code with comprehensive documentation, citation data, input variables, and reviews spreadsheets is provided, making this review replicable, open-source, and free from human errors in selecting studies. The reviewed literature is categorized based on type of prediction systems to demonstrate the evolution of techniques and research gaps. The reviewed literature is 7 % statistical, 9% machine learning, 23% deep learning, 20% hybrid, 25% combination of machine learning and deep learning, and 14% studies explore multiple categories of techniques. This review provides detailed information on prediction techniques, competitor techniques, performance metrics, input variables, data timing, and research gap to enable researchers to create prediction systems per technique category. The review showed that stock trading data is most used and collected from Yahoo! Finance. Studies showed that sentiment data improved stock prediction, and most papers used tweets from Twitter. Most of the reviewed studies showed significant improvements in predictions to previous systems.KeywordsExpert systemsHybrid learningStock marketForecastingDeep learningArtificial intelligenceMachine learning
... Such algorithms deal with the construction of machines that move automatically by gaining experience, the formation of these algorithms with low computational costs, the design of new algorithms and the usability of big data have made progress in recent years. Since artificial intelligence has a very wide usage area, studies on artificial intelligence can be found in nearly every subject when reviewing the literature [16][17][18][19][20][21][22][23][24][25][26][27][28]. Due to artificial intelligence, computers can be programmed to perform specific tasks or process [29], desired classifications can be made [30][31][32][33], models can be designed, and these models can make predictions about the future [21,34], based on previous experiences or dataset presented as examples [35]. ...
Human activities are linked to atmospheric pollution and are affected by economic development. Ground-level ozone has become an important and harmful pollutant for many countries, adversely affecting public health. As there is a limited number of on-site measurements, alternative methods are required to accurately estimate ozone concentrations. In this study, a database containing annual average concentrations of CO2, N2O, CO, NOx, SOx, and O3, covering the years 2008-2018 for ten countries in Europe, was created. Ten different artificial intelligence regression methods were developed to predict the O3 concentration using these variables. The predictive performance of the developed artificial intelligence models was compared using the coefficient of determination, mean absolute error, root mean square error, and relative absolute error criteria. Experimental results show that the Bagging-MLP method has a better predictive performance than other models in ozone concentration estimation, with an R 2 value of 0.9994, mean absolute error of 24.67, root mean square error of 33.85, and relative absolute error of 2.9%. This study shows that the O3 concentration can be estimated very close to the actual value by using the Bagging-MLP method, one of the artificial intelligence methods. Bagging-MLP Yöntemiyle Troposferik Ozon Konsantrasyonunun Tahmini ÖZ İnsan faaliyetleri atmosfer kirliliği ile bağlantılıdır ve ekonomik gelişmelerden etkilenir. Yer seviyesindeki ozon birçok ülke için önemli ve zararlı bir kirletici haline gelmiş olup halk sağlığını olumsuz etkiler. Yerinde yapılan ölçümlerin sınırlı sayıda olmasından dolayı, ozon konsantrasyonlarını doğru bir şekilde tahmin etmek için alternatif yöntemlere ihtiyaç vardır. Bu çalışmada, Avrupa'da on ülkede 2008-2018 yıllarını kapsayan CO2, N2O, CO, NOx, SOx, ve O3 yıllık ortalama konsantrasyonlarını içeren bir veritabanı oluşturuldu. Bu değişkenleri kullanarak O3 konsantrasyonunu tahmin etmek için on farklı yapay zeka regresyon yöntemi geliştirildi. Geliştirilen yapay zeka modellerinin tahmin performansı, determinasyon katsayısı, ortalama mutlak hata, kök ortalama karesel hata ve göreceli mutlak hata ölçütleri kullanılarak karşılaştırıldı. Deneysel sonuçlar, Bagging-MLP yönteminin diğer modellere göre ozon konsantrasyonu tahmininde daha iyi bir performansa sahip olduğunu, R 2 değeri 0.9994, ortalama mutlak hata 24.67, kök ortalama karesel hata 33.85 ve göreceli mutlak hata ise %2.9 olarak ortaya koydu. Bu çalışma, yapay zeka yöntemlerinden olan Bagging-MLP yöntemi kullanılarak O3 konsantrasyonunun gerçek değere oldukça yakın bir şekilde tahmin edilebileceğini göstermektedir.
... Artificial neural networks and Genetic Algorithms among others are examples of tools and methodologies used in artificial intelligence research field to anticipate stock market movements. Many studies have looked into the benefits of using such approaches including and are not limited to (Nevasalmi 2020;Patel et al. 2015;Khan et al. 2020 Hassanat et al. 2015bHassanat et al. , 2015cTarawneh et al. 2020a;Hassanat and Tarawneh 2016), Software engineering (Salman et al. 2018;Eyal Salman 2017;Eyal Salman et al. 2015), Internet of things (Mnasri et al. 2014(Mnasri et al. , 2015(Mnasri et al. , 2017a(Mnasri et al. , 2017b(Mnasri et al. , 2018a(Mnasri et al. , 2018b(Mnasri et al. , 2019(Mnasri et al. , 2020Abdallah et al. 2020aAbdallah et al. , 2020bTlili et al. 2021), Computer vision (AlTarawneh et al. 2017Alqatawneh et al. 2019;Tarawneh et al. 2018Tarawneh et al. , 2019aTarawneh et al. , 2019bTarawneh et al. , 2020bAl-Btoush et al. 2019;Hassanat et al. , 2017aHassanat et al. , 2017bHassanat et al. , 2018aHassanat and Tarawneh 2016;Hassanat 2018e), Game theory (De Voogt et al. 2017;Hassanat et al. 2018b), Big data classification (Hassanat 2018a(Hassanat , 2018b(Hassanat , 2018c(Hassanat , 2018d Hassanat et al. 2022). Security is a field that can benefit from machine learning techniques. ...
One of the most difficult problems analysts and decision-makers may face is how to improve the forecasting and predicting of financial time series. However, several efforts were made to develop more accurate and reliable forecasting methods. The main purpose of this study is to use technical analysis methods to forecast Jordanian insurance companies and accordingly examine their performance during the COVID-19 pandemic. Several experiments were conducted on the daily stock prices of ten insurance companies, collected by the Amman Stock Exchange, to evaluate the selected technical analysis methods. The experimental results show that the non-parametric Exponential Decay Weighted Average (EDWA) has higher forecasting capabilities than some of the more popular forecasting strategies, such as Simple Moving Average, Weighted Moving Average, and Exponential Smoothing. As a result, we show that using EDWA to forecast the share price of insurance companies in Jordan is good practice. From a technical analysis perspective, our research also shows that the pandemic had different effects on different Jordanian insurance companies.
... The usage of ML algorithms allows improving accuracy and overcome limitations of common econometrical models (Rossi 2018). Moreover, the ML algorithms outperform the benchmark buy-and-hold strategy in the real-life simulations and the gradient boosting machine performs the best from the perspective of the statistical and economic evaluation criteria (Nevasalmi (2020)). ...
The international rankings of universities have a significant impact on the perception of academics, students, governments, and businesses toward the universities. The aim of the study is to identify country-specific factors that are thought to influence academic performance and to reveal how the country-specific factors should be developed for universities to increase their success. For this purpose, the University Ranking by Academic Performance index for universities in 103 countries covering the period 2013–2019 was taken as the dependent variable to represent academic performance. The index value, which is used as the dependent variable, was constituted by taking the average of the countries to which entered the ranking universities belong. The country-specific factors were taken as the political stability and absence of violence/terrorism, the rule of law, the freedom of the press, the economic freedom index, the university-industry collaboration in research & development index, and the gross domestic product per capita. The factors affecting the academic performance of universities were analyzed with spatial panel data methods, and the findings revealed that the rule of law, the university-industry collaboration, and the GDP per capita increase academic performance.
... The usage of ML algorithms allows improving accuracy and overcome limitations of common econometrical models [29]. Moreover, the ML algorithms outperform the benchmark buy-and-hold strategy in the real-life simulations and the gradient boosting machine performs the best from the perspective of the statistical and economic evaluation criteria [27]. ...
In the modern world, online social and news media significantly impact society, economy and financial markets. In this chapter, we compared the predictive performance of financial econometrics and machine learning and deep learning methods for the returns of the stocks of the S&P 100 index. The analysis is enriched by using COVID-19-related news sentiments data collected for a period of 10 months. We analysed the performance of each model and found the best algorithm for such types of predictions. For the sample we analysed, our results indicate that the autoregressive–moving-average model with exogenous variables (ARMAX) has a comparable predictive performance to the machine and deep learning models, only outperformed by the extreme gradient boosted trees (XGBoost) approach. This result holds both in the training and testing datasets.
