جستجوی مقالات مرتبط با کلیدواژه "ماشین بردار پشتیبان" در نشریات گروه "مالی"

In recent years, data mining, particularly the support vector machine, has gained considerable interest among investors, managers, and researchers as an effective means of bankruptcy prediction. However, studies indicate that it is highly sensitive to the selection of parameters and input variables. Hence, the aim of this research is to improve bankruptcy prediction accuracy by combining an advanced support vector machine model with the k-nearest neighbor approach to eliminate erroneous entries. To achieve this, first, by using five financial ratios: current ratio, net profit margin, debt ratio, return on assets, and return of investment from 150 companies listed on the Tehran Stock Exchange during the 10-year period (2010-2019), and k-nearest neighbor algorithm, the training data will be refined. Then, relying on a support vector machine based on classification penalty, a prediction model will be constructed. The parameters will be estimated, and its validity will be assessed using test data. Finally, a comparison will be made between the outcomes of the proposed model and traditional models.The findings demonstrate that the combination of the k-nearest neighbor models and support vector machine reduces the overall prediction error, and the penalty coefficients of the support vector machine exhibit a high level of statistical significance.

This article aims to forecast Iran Cooperative Development Bank's profit/loss using a two-stage collective learning method. Employing machine learning for profit and loss prediction is a novel approach to numerical computations, aligning with the article's goal of leveraging big data. Hence, we employ a two-stage collective learning method utilizing Support Vector Machines, Decision Trees, and Weighted Averaging models for learning, testing, and predicting the profit/loss of the Cooperative Development Bank.

In the initial stage of machine learning, Support Vector Machines, and Decision Trees serve as the base models, while the second stage employs a weighted averaging approach. The combination of base learning models was utilized to initially test the prediction process on the performance of 35 branches of the Cooperative Development Bank nationwide from 2012 to 2021. The two-stage machine learning method relies on the utilization of 12 variables grouped into 5 factors for the task. These variables encompass interest-free loans and short-term deposits, deposit interest expenses, total customer deposits, income from facilities and investments, common income, cash balances, administrative expenses, long-term deposits, non-common income, asset depreciation expenses, and the bank's size. Relevant data for the study was obtained from the financial statements of 35 branches of the Cooperative Development Bank spanning the years 2012 to 2021. To minimize prediction errors and facilitate ratio comparisons, all indicators mentioned were adjusted relative to the total value of the bank's assets. Furthermore, to minimize prediction errors and enhance the comparability of ratios, all the mentioned indicators were adjusted in proportion to the total value of the bank's assets. In the initial stage of applying this method, two machine learning models - Support Vector Machines and Decision Trees - were employed, followed by a weighted averaging approach in the second stage.

This article contrasts linear regression with machine learning approaches for predicting the Cooperative Development Bank's actual profit/loss. The results reveal notably high-performance accuracy, evidenced by an MAE metric of 5.66 and an MSE metric of 620.34. Additionally, the correlation between training data and predictions from the two-stage collective machine learning stands at 0.9977. Following this method's performance assessment, it is subsequently employed to predict the Cooperative Development Bank's profit/loss for the years 2022 to 2027.

The results, showcasing the high efficiency of the two-stage collective machine learning method, suggest that managers can employ these approaches for profit/loss prediction in banks. Based on this method, and under normal conditions without abnormal or non-normal circumstances, the obtained results indicate a prospective decrease in the accumulated losses of the Cooperative Development Bank in future years, leading to an ultimate increase in profits by the year 1405. The results of data analysis reveal that the average ratio of the net profit or loss of the bank to its total assets has been computed in a manner that reflects the average profitability of the bank branches over the research period.

Earnings and earnings per share information are metrics that are considered important by many users; Therefore, companies try to attract investors with the most accurate forecast of earnings per share. On the other hand, despite the various methods of forecasting earnings, accurate forecasting of earnings per share in the financial field is not easy and most researchers are trying to determine the best way to forecast earnings; Therefore, the main purpose of this study is to investigate the accuracy of support vector machine based on genetic algorithm compared to common linear methods in forecasting earnings per share. For this purpose, samples consisting of 100 companies listed on the Tehran Stock Exchange during the years 2008-2019 have been studied. In order to achieve the objectives of the research, first by studying previous research in the field of earnings forecasting, 14 financial ratios affecting earnings forecasting have been selected. Then, in order to provide a model for predicting the profitability of companies, a combined model of support vector machine based on genetic algorithm, support vector machine and linear regression is compared. The results showed that the hybrid model of support vector based on genetic algorithm is much better in predicting the trend of earnings per share and has a higher accuracy compared to the model of support vector based on kernel functions and linear regression method. Thus, with the development of the support vector machine model based on the genetic algorithm, the model training error is reduced to 0.036 and the accuracy of the model is increased up to 75%.

Choosing the right portfolio is always one of the most important issues for investors. The price trend is predicted using technical analysis or basic analysis. Technical analysis focuses on market performance, while the focus of fundamental analysis is on the mechanism of supply and demand, and these changes prices. The existence of a solution to predict growth or decrease in stocks has been studied as a basic need in this study. In the present study, with the help of a monitoring dataset, a solution based on Raff collection algorithms and hierarchical analysis to reduce the feature and decision tree algorithms, backup vector machine, and business network have been used for prediction. This proposed solution has been implemented using language and compared with different solutions, and the research results have shown that the proposed method with 80% accuracy of prediction and 20 errors in prediction has the highest accuracy and the lowest error rate among the methods compared.

Determining the appropriate strategy for buying or selling in the foreign exchange market is very important for companies to cover exchange rate fluctuations against the national currency. This study proposes a new approach based on genetic algorithms and support vector machines for trading in the foreign exchange market.In this research, a new algorithm with the ability to generate technical rules for investment based on forecast certainty is presented. For prediction, a combination of the Combined Support Vector Machine (HSVM) algorithm for classifying the market into three different classes (uptrend, downtrend, sideway) and a dynamic genetic algorithm for optimizing trading rules based on several technical indicators Different has been used. Rials-dollar pair data is used as training and test data for the period between 1392 and 1398. The proposed architecture for machine learning, as well as the implementation and study of the proposed trading system are fully described. The research shows promising results during the test period in which the return on investment was 129%.

One of the most important issues in the field of financial management is that investors can distinguish favorable investment opportunities from unfavorable ones. One way to help investors is to anticipate the financial recovery (exit from helplessness) of companies with financial distress. Therefore, this study intends to provide a model for predicting financial recovery using the support vector machine algorithm for companies listed on the Tehran Stock Exchange. To achieve this goal, 54 financial variables were determined using the Lars feature selection algorithm and to test the accuracy of the results of the proposed model, the support vector learning algorithm was used. For this purpose, in the period of 2001-2018, the information of 167 helpless companies that were out of financial helplessness and revived was extracted. The research findings show that the research model accurately predicts the recovery time of the financially helpless company from financial distress with 74% accuracy.

The purpose of this article is to predict impending financial distress of the listed companies on Tehran Stock Exchange (TSE) and Iran Fara Bourse (IFB) using a wide range of features including accrual accounting variables, cash-based accounting variables, market-based variables, corporate governance mechanisms, and macroeconomic indicators. The final sample includes 421 firms leading to 3,670 firm-year observations. The prepared data, was then split into a train and test data set using a 70/30 ratio. In this research, various data pre-processing machine learning techniques i.e., Z-score standardization, one-hot encoding, stratified K-fold validation combined with feature engineering are applied to improve classifier performance. Stratified K-fold cross validation method, (with k = 5) was used for estimation of model prediction performance during training phase. During the training phase, hyper-parameter tuning of a model was carried out using a grid-search. Furthermore, a cost-sensitive meta-learning approach in conjunction with the proposed imbalance-oriented metric i.e., F1 score were used to overcome the extreme class imbalance issue. Based on the experimental results, the tuned Support Vector Machine (SVM) model achieved a f1-score, MCC, recall and precision of respectively, 55%, 56%, 78% and 43% on the training set. Finally, the proposed model was tested on the hold-out test set which resulted in a f1-score, MCC, recall and precision of 50%, 50%, 68% and 40%, respectively.

Today, the Forex market is the largest financial market in the world. Determining the right strategy for buying or selling in the Forex market is based on predicting the price trend. Therefore, to choose a suitable strategy in Forex, complex meta-heuristic models are used. In this research, by predicting the market trend and based on trading rules based on fuzzy technical indicators, a new method for investing in the Forex market is presented. For forecasting, a combination of hyper support vector machine (HSVM) algorithm is used and for market classification in three different classes (uptrend, downtrend, sideway) and a dynamic genetic algorithm is used to optimize trading rules. Five fuzzy technical indicators have been used to determine the trading rules. Euro-dollar pair data is used as daily training and test data for a daily period between 2010 and 2019. The results obtained compared to traditional methods have had promising results.

One of the most important issues faced by banks and financial institutions is the issue of credit risk. The significant amount of deferred bank claims around the world indicates the importance of this issue and the need to pay attention to it. So far, many efforts have been made to provide an effective model for evaluating and classification credit applicants as accurately as possible. In this regard, the present study attempts to provide a new approach for assessing the credit risk of bank customers. The support vector machine(SVM) method is combined as the main classifier of banking customers, with a feature selection method called the Binary Ant Colony Optimization Algorithm(BACO-SVM). In order to demonstrate the effectiveness of the proposed method, we used data from 85 companies from legal recipients of facilities of an Iranian bank in a 5 year interval (1393-1893) along with 16 characteristics related to each of them. The results of the BACO-SVM method have been compared with the PSO-SVM, GA-SVM, and SVM method. The results of the research indicated that BACO-SVM model has better performance in assessing credit risk rather than other methods. As the result, using the BACO-SVM method, we classify customers into two groups of good and bad account customers. Finally, in order to increase the flexibility in decision making, we will rank our good account customers with the VIKOR method. This rating will lead to a more accurate assessment of the credit risk situation of good account applicants.

Nowadays, financial distress prediction is one of the most important research issues in the field of risk management that has always been interesting to banks, companies, corporations, managers and investors. The main objective of this study is to develop a high performance predictive model and to compare the results with other commonly used models in financial distress prediction For this purpose, sequential floating forward selection that is considered as the generalized form of sequential forward selection method and as one of the wrapper methods, and sequential forward selection methodin combination with support vector machine were used. These models are combined models of feature selection and classifier. Logistic regression model which is a statistical classification models, has also been used in the present study. After reviewing the important financial ratios, 29 financial ratios that were mostly used in previous researches were chosen. Paired T-test results showed thatwith a 95% confidence level. The proposed model provides higher accuracy than other models used in this study. Results showed that the proposed model of this research has significantly better performance in predicting financial distress than the sequential forward selection method and Logistic regression model in one year, two years and three years before financial distress

Stock return forecasting is one of the most important question for investing in Stock markets. Because of the effects of policy, economic, etc., we need moderns and intelligent models to forecast the returns.
The main idea in this research is classifying the stocks into high and low return groups, for this purpose support vector machine (SVM) was used. To elect the best variables for models we used sequential feature selection and in order to evaluate the accuracy of SVM we do the same forecasting with diagonal quadratic discriminant analysis (DQDA). By using paired t-test, we conclude that models have no significant difference.
Equal weighted portfolios were created for each models with and without feature selection also, we used posterior probability to weight the portfolio of DQDA with feature selection. The returns were calculated for each portfolio during the years 1388-1391. The simulating results are satisfying and all portfolios’ returns are better than market portfolio.

Financial distress prediction (FDP) is a great important subject that has always been interesting to researchers, financial institutions and banks. Tough many works have been done in this area, but use of combined approach of feature selection and classifier is an issue that has attracted researcher's attention just in recent years. In this paper, four well-known kinds of SVM that each of them has it's own kernel function including: linear, polynomial, radial and sigmoid have been introduced as the main classifiers of our proposed approach. These four methods have been integrated with genetic algorithm (GA) as a wrapper feature selection approach as well as three techniques of filtering feature selection approach called: principle component analysis (PCA), information gain and relief. Brought results indicated that genetic algorithm outperformed the other feature selection techniques in it's combination with SVM methods. Furthermore, implemented hypothesis test implied that there was no significance level among GA-SVM (linear), GA-SVM (radial), GA-SVM (polynomial) and GA-SVM (sigmoid) techniques with confidence level of %95.

In this study we focus on developing a stock trend prediction model based on a modified version of support vector machine, named volume weighted support vector machine, along with a hybrid feature selection method named FSSFS method. In order to evaluate the prediction accuracy of this model we compare the VW-SVM classifier with plain support vector machine along with three commonly used feature selection methods including Information gain, Symmetrical uncertainty and correlation-based feature selection, via paired t-test. As the model input, we use several technical indicators and statistical measures, calculated for 10 stocks. The results show that the VW-SVM, combined with the hybrid feature selection method, significantly outperforms plain SVM model to the problem of stock trend prediction. In addition our experimental result show that VW-SVM combined with F-SSFS has the highest level of accuracies and generalization performance in comparison with the other three feature selection methods. With these results, we claim that VW-SVM combined with F-SSFS can serve as a promising addition to the existing stock trend prediction.

The trend of the stock price index, has taken as one of the investment criteria consistently. Because of the two components of nonlinear and time series price index volatility, in this study, a new hybrid model presented that can predict move and change of these two components of the trend of the index with the highest accuracy. In this model, at the first by using wavelet transform the index time series splits into six separate time series index which represent the characteristics of nonlinear and volatility of index. Then the derived time series with nonlinear behavior by combining the support vector machine and particle swarm optimization (SVM-PSO) and time series behavior based on index volatility by using GJR models predicted and then by accumulating results of two nonlinear and volatility of price index prediction component, price index time series estimates. The results show that the proposed hybrid model, in comparison to other forecasting methods, has fewer errors and higher accuracy.

In this study, a prediction model based on support vector machines (SVM) improved by introducing a volume weighted penalty function to the model was introduced to increase the accuracy of forecasting short term trends on the stock market to develop the optimal trading strategy. Along with VW-SVM classifier, a hybrid feature selection method was used that consisted of F-score as the filter part and supported Sequential forward selection as the wrapper part, to select the optimal feature subset. In order to verify the capability of the proposed model in successfully predicting short term trends, a trading strategy was developed. The model input included several technical indicators and statistical measures that were calculated for chosen 10 stocks from Tehran Stock Exchange. The results show that the VW-SVM, combined with the hybrid feature selection method, significantly increases the profitability of the proposed strategy compared to rival strategies, in terms of both overall rate of return and the maximum draw down during trading period.

Design and implementation of credit rating model in the banking system plays an important role in enhancing the efficiency of resource allocation is to target customers. In this research aims to develop a model for evaluating the credit risk of the bank's corporate clients have been used Support Vector Machine (SVM) and Genetic Algorithms (GA). Therefore, a study has been on the financial variables of 282 companies during the years 2007 to 2010, have received loans from TEJARAT bank. In this research, to optimize the input of support vector machine is used of genetic algorithms. The power of the genetic algorithm to select the optimum points, always provides confidence that the optimal-made for the proposed going to be higher optimum points. In the hybrid model GA-SVM, genetic algorithm optimizes SVM model inputs the data.
Research findings show GA-SVM hybrid model performed better than the SVM model in the identifying good customer accounts and bad customer accounts and credit risk prediction.

According to recent developments of predicting methods in financial markets، and since the stock price is one of the most important factors for investment decision-making، and its prediction can play an important role in this field، the aim of this study is to provide a model to predict the stock price movement with high accuracy. Accordingly، a hybrid model for predicting the stock price movement using Support Vector Machine (SVM) based on genetic algorithms is presented. Thirty companies from the top 50 companiesin Tehran Stock Exchange in 2011 are selected as sample. Then، for each company، 44 variables have been calculated. These variables are the inputs of the hybrid model and are optimized using genetic algorithm. The results show that the hybrid model of Support Vector Machine based on genetic algorithms has better performance in predicting the stock price movement and it has a higher accuracy compared with the simple Support Vector Machine.

One of the main subjects of financial management is risk management. Risk management involves recognizing، measuring and monitoring risk. So measuring risk is very important part of the risk management. One of the most recognized and applied way of measuring risk، is evaluating value at risk that is the main subject of this research. In this research، we forecast volatility of TEPIX index and TSE-50 index، using the hybrid model of support vector machine based and the GARCH، then we calculate Value at Risk by Variance-Covariance approach and finally we compare its result with the traditional models including: Risk Metrics، GARCH and EGARCH by LOPEZ’s back testing and back testing based of expected shortfall. The result of this research has shown that the hybrid model significantly outperform the competing models.

Ability to predict financial distress as one of the areas of risk management has various social and individual aspects. The aim of this study is to improve predicting financial distress process relying on two important processes. The first part of this article focuses on the financial distress predictor variables and then the two major models of financial distress prediction have been emphasized. For the goal, 20 financial ratios and efficiency of corporate as non-financial predictive variable, entering the two predictive models also differ in nature mean support vector machines and multiple discriminate analyses. The efficiency was computed by DEA. In order to have a real and effective estimate of predictive variables, by using feature selection method among variables, 10 financial ratios beside efficiency have chosen. Also to avoid of over fitting, cross-validation method with 12 v-folds has been conducted. As well as to compute the influence of efficiency on prediction accuracy, the prediction models were conducted by with and without efficiency respectively. Because of lack of the significant difference in the results of mentioned models, all hypotheses were rejected.