Wavelet Neural Networks for Volatility Forecasting. Comparative Analysis with Stochastic Models

Transcom Proceedings 2015: 11-th European Conference of Young Researchers and Scientists: Section 8: Natural Sciences (Applied Mathematics). Social Sciences 2015
Andrejs Pučkovs

This paper provides volatility forecasting models by using wavelet analysis, variance analysis and NARX architecture neural networks. Implementation of volatility forecasting is applicable for financial market players: institutional investors (banks, insurance companies, private and state pension funds) as well as private investors for volatility forecasting in financial time series. Developed models are compared with worldwide used stochastic models like GARCH(P,Q), EGARCH(P,Q) and GJR(P,Q). According to research results developed models provide better forecasting results in terms than conditional variance models

Atslēgas vārdi
Nonlinear Autoregressive with exogenous inputs (NARX) neural networks, wavelet analysis, wavelet neural networks, direct discrete wavelet transform, volatility, historical volatility, implied volatility, volatility index (VIX), variance, the north-east volatility wind' effect, stochastic models, conditional variance models, GARCH(P,Q), EGARCH(P,Q) , GJR(P,Q), time series, forecasting.
Hipersaite
http://www.transcom-conference.com/uploads/archive/2015/S08_2015_Proceedings.pdf#page=64

Pučkovs, A. Wavelet Neural Networks for Volatility Forecasting. Comparative Analysis with Stochastic Models. No: Transcom Proceedings 2015: 11-th European Conference of Young Researchers and Scientists: Section 8: Natural Sciences (Applied Mathematics). Social Sciences, Slovākija, Žilina, 22.-24. jūnijs, 2015. Žilina: University of Zilina, 2015, 63.-68.lpp. ISBN 978-80-554-1050-0. ISSN 1339-9799. e-ISSN 1339-9829.

Publikācijas valoda
English (en)