Skip to main content

LUP Student Papers

LUND UNIVERSITY LIBRARIES

Return Models and Covariance Matrices

Xie, Xiaolei LU (2014) FYSM60 20141
Department of Physics
Mathematical Physics
Abstract
Return models and covariance matrices of return series have been studied. In particular, GARCH and SV models are compared with respect to their forecasting accuracy when applied to intraday return series. SV models are found to be considerably more accurate and more consistent in accuracy in forecasting.

Covariance matrices formed from Gaussian and GARCH return series, and in particular, return series auto-correlated as an AR(1) process, have been studied. In the case of Gaussian returns, the largest eigenvalue is found to approximately follow a gamma distribution also when the returns are auto-correlated. Expressions relating the mean and the variance of the asymptotic Gaussian distribution of the matrix elements are derived. In the... (More)
Return models and covariance matrices of return series have been studied. In particular, GARCH and SV models are compared with respect to their forecasting accuracy when applied to intraday return series. SV models are found to be considerably more accurate and more consistent in accuracy in forecasting.

Covariance matrices formed from Gaussian and GARCH return series, and in particular, return series auto-correlated as an AR(1) process, have been studied. In the case of Gaussian returns, the largest eigenvalue is found to approximately follow a gamma distribution also when the returns are auto-correlated. Expressions relating the mean and the variance of the asymptotic Gaussian distribution of the matrix elements are derived. In the case of GARCH returns, both the largest and the smallest eigenvalues of the covariance matrix are seen to increase with increasing auto-correlation. The matrix elements are found to follow Levy distributions with different Levy indexes for the diagonal and the non-diagonal elements.

Localization of eigenvectors of covariance matrices of returns from GARCH processes has been investigated. It is found that the localization is reduced as the auto-correlation is increased. Quantitatively, the number of localized eigenvectors
decreases approximately as a quadratic function with the auto-correlation strength, i.e. the autoregressive coefficient of the AR(1) process. (Less)
Popular Abstract
Can we predict the movements of the financial market? The direct answer is "no", but there is more to be said. One cannot predict the price movement in an efficient market, but one can indeed predict how "volatile" the prices are (volatility). This does not lead to profits, but is useful for risk management --- if one pursues a profit, one must take a risk. Knowing and quantitatively managing the risks associated with investments, one often makes wiser decisions.

This report entails, first of all, a comparison of return models in the context of intraday returns. 6 return series are studied: Nordea, Volvo and Ericsson in the time scales of 15 and 30 minutes. Stochastic Volatility (SV) models and GARCH models are fitted to these series... (More)
Can we predict the movements of the financial market? The direct answer is "no", but there is more to be said. One cannot predict the price movement in an efficient market, but one can indeed predict how "volatile" the prices are (volatility). This does not lead to profits, but is useful for risk management --- if one pursues a profit, one must take a risk. Knowing and quantitatively managing the risks associated with investments, one often makes wiser decisions.

This report entails, first of all, a comparison of return models in the context of intraday returns. 6 return series are studied: Nordea, Volvo and Ericsson in the time scales of 15 and 30 minutes. Stochastic Volatility (SV) models and GARCH models are fitted to these series and the resulting volatility forecasts are compared. SV models are found to be more accurate in most cases.

Covariance matrices, which give the correlation between different companies, are also studied. In particular, the eigenvectors and eigenvalues of a covariance matrix, which respectively represent a set of driving factors of the companies' shares and the variances of these factors, are studied. It is found that when the return series are auto-correlated, i.e. when observations at different times are correlated, the eigenvectors are delocalized --- their expansion as a linear combination of the basis vectors involve more appreciable coefficients. (Less)
Please use this url to cite or link to this publication:
author
Xie, Xiaolei LU
supervisor
organization
course
FYSM60 20141
year
type
H2 - Master's Degree (Two Years)
subject
keywords
returns, stochastic volatility, GARCH, covariance matrix, random matrix, spectral distribution
language
English
id
4519670
date added to LUP
2014-06-27 10:30:57
date last changed
2015-12-14 13:32:33
@misc{4519670,
  abstract     = {{Return models and covariance matrices of return series have been studied. In particular, GARCH and SV models are compared with respect to their forecasting accuracy when applied to intraday return series. SV models are found to be considerably more accurate and more consistent in accuracy in forecasting.

Covariance matrices formed from Gaussian and GARCH return series, and in particular, return series auto-correlated as an AR(1) process, have been studied. In the case of Gaussian returns, the largest eigenvalue is found to approximately follow a gamma distribution also when the returns are auto-correlated. Expressions relating the mean and the variance of the asymptotic Gaussian distribution of the matrix elements are derived. In the case of GARCH returns, both the largest and the smallest eigenvalues of the covariance matrix are seen to increase with increasing auto-correlation. The matrix elements are found to follow Levy distributions with different Levy indexes for the diagonal and the non-diagonal elements.

Localization of eigenvectors of covariance matrices of returns from GARCH processes has been investigated. It is found that the localization is reduced as the auto-correlation is increased. Quantitatively, the number of localized eigenvectors 
decreases approximately as a quadratic function with the auto-correlation strength, i.e. the autoregressive coefficient of the AR(1) process.}},
  author       = {{Xie, Xiaolei}},
  language     = {{eng}},
  note         = {{Student Paper}},
  title        = {{Return Models and Covariance Matrices}},
  year         = {{2014}},
}