Towards Efficient Risk Quantification-Using GPUs and Variance Reduction Technique
Universite Nice Sophia Antipolis
hal-00932233, (20 January 2014)
@article{tan2013towards,
title={Towards Efficient Risk Quantification-Using GPUs and Variance Reduction Technique},
author={Tan, Shih Hau},
year={2013}
}
Value-at-Risk (VaR) provides information about global risk in trading. The request for high speed calculation about VaR is rising because financial institutions need to measure the risk in real time. Researchers in HPC also recently turned their attention on this kind of demanding applications. In this master thesis, we introduce two complementary and different strategies to improve VaR calculation: one is directly coming from financial mathematics, the other pertains to take advantage of high performance recently available computing devices: GPUs. Our aim is to study the potential of these two approaches on well chosen examples in order to evaluate how much computing time we can spare. Eventually, we discuss alternate approaches worth to be studied in future works.
January 30, 2014 by hgpu
