Various factors make stock market forecasting difficult and arduous. Single-task learning models fail to achieve good results because they ignore the correlation between multiple related tasks. Multitask learning methods can capture the cross-correlation among subtasks and achieve a satisfactory learning effect by training all tasks simultaneously. With this motivation, we assume that the related tasks are close enough to share a common model whereas having their own independent models. Based on this hypothesis, we propose a multitask learning least squares support vector regression (MTL-LS-SVR) algorithm, and an extension, EMTL-LS-SVR. Theoretical analysis shows that these models can be converted to linear systems. A Krylov-Cholesky algorithm is introduced to determine the optimal solutions of the models. We tested the proposed models by applying them to forecasts of the Chinese stock market index trend and the stock prices of five stated-owned banks. The experimental results demonstrate their validity.

Multitask learning can capture the relationship among tasks and achieve a better performance by training all tasks simultaneously, which overcomes the shortcomings of the single-task learning method. Asymmetric squared loss function can improve the generalization performance of the model by adjusting the asymmetric parameters. In this paper, we first replace the squared loss function with asymmetric squared loss function. Then, we make two multitask assumptions on the relatedness among subtasks, and propose two multitask learning proximal support vector machine algorithms by using asymmetric squared loss function, named as multitask learning asymmetric proximal support vector machine (MTL-a-PSVM) and extension of multitask learning asymmetric proximal support vector machine(EMTL-a-PSVM) respectively. Both multitask algorithms can obtain the optimal solution by solving the corresponding quadratic programming problem. Besides, two special cases corresponding to MTL-a-PSVM and EMTL-a-PSVM are proposed by analyzing the asymmetric squared loss function. Experimental analysis on the two groups of classification datasets demonstrates the effectiveness and superiority of our proposed multitask algorithms.