Timing Selection And Stock Selection Based On Support Vector Machines

Sample number in traditional statistics is big, so statistics assume the sample number is infinity. However, in many practical cases, samples are small samples, the samples we can get are several, or just more than ten number. Most of existing methods based on traditional statistical theory may not work well in the situation of small samples. Statistical Learning Theory (SLT) is a statistical theory for finite samples that fit these small-sample cases. Support Vector Machines (SVMs) is a new learning machine built on VC (Vapnik-Chervonenkis) dimension and Structural Risk Minimum principle of SLT. SVMs do well in solving small-samples problems in practice.Financial engineering often meets small-sample problems where the number of samples is not much bigger than the dimension of the sample, for example, when we select investment targets using financial information of listed companies, there are maybe a little more than ten annual reports of listed companies, however there are much more financial indicates(more than thirty) can be used. SVMs fit these situations. The thesis investigates the applications of Support Vector...