Financial time series forecasting using twin support vector regression

Abstract

Financial time series forecasting is a crucial measure for improving and making more robust financial decisions throughout the world. Noisy data and non-stationarity information are the two key factors in financial time series prediction. This paper proposes twin support vector regression for financial time series prediction to deal with noisy data and nonstationary information. Various interesting financial time series datasets across a wide range of industries, such as information technology, the stock market, the banking sector, and the oil and petroleum sector, are used for numerical experiments. Further, to test the accuracy of the prediction of the time series, the root mean squared error and the standard deviation are computed, which clearly indicate the usefulness and applicability of the proposed method. The twin support vector regression is computationally faster than other standard support vector regression on the given 44 datasets.

Dynamics of Heteroscedasticity Modelling and Forecasting of Tax Revenue in a Developing Economy: A Review

“…Grigoryan [69] applied Independent Component Analysis (ICA) and Support Vector Machines (SVM) to examine the stock market prediction based method their study shows that the hybrid of SVM and feature selection technique methods are effective technique for the stock market forecast. Besides, the application of the pre-processing technique on the data yields a better result.…”

“…On their study attention was concentrated on the model performance ability living data and forecasting accuracy unpredicted. [65], [66], [67], [68], [69], [70] [71], [73], [76], [79], [80], [81] Linear model 2 [62], [72] Non-linear model 13 [65], [66], [68], [69], [70], [71], [72], [73], [76], [79], [80], [81], [82], Heteroscedasticity 9 [69], [70], [71], [73], [76], [79], [80], [81] Table 2 reveals the overview of the studied literature on stock exchange index. With over 80% of the literatures on stock exchange index utilized nonlinear model for forecasting stock price index with consideration for heteroscedasticity in the data.…”

A Holistic Auto-Configurable Ensemble Machine Learning Strategy for Financial Trading

“…The work in [38] used Principal Component Analysis to reduce the dimensionality of the data, Discrete Wavelet Transform to reduce noise, and an optimized Extreme Gradient Boosting to trade in financial markets. The work in [39] validated an extension of Support Vector Regression, called Twin Support Vector Regression, for financial time series forecasting. The work in [40] proposed a novel fuzzy rule transfer mechanism for constructing fuzzy inference neural networks to perform two-class classification, such as what happens in financial forecasting (e.g., buy or sell).…”

“…In addition, we performed the future market predictions by using single predictors (i.e., GB, SVM, and RF), configuring their default hyper-parameters according to some common values in the literature: 40% of the IS dataset as walk size, of which 75% is used as training set and the remaining 25% as validation set with 5 day-lags [15,16,[73][74][75]. Finally, we also used a recent approach to perform trading [39], which we call TSVR in the remaining of this paper. For this approach, we used both the linear (LIN) and nonlinear (NONLIN) kernel.…”

“…For this approach, we used both the linear (LIN) and nonlinear (NONLIN) kernel. As described in in [39], we have used 10-fold cross validation in the training data to find the kernel parameters that yielded the best mean squared error in all markets. As this approach is proposing to predict the closing price (regression problem), we mapped the problem consistently with ours and changed the output so that for each day we have either a long or short operation.…”