Extreme events: blending PCA and ICA [35]

Go to: Summary | Previous | Next   
Bullet points include: Possible ways of adjusting for recent past time-varying volatility include. Longitudinal: adjust each series in isolation by a different (time-varying) factor dependent its recent past volatility, or Cross-sectional: adjust every series by the same (time-varying) factor dependent on the average spread of returns across the sectors in the recent past. Using contemporaneous data, such as implied volatilities and correlations (not analysed further here, discussed in more detail in “Market Consistency”). E.g. use rolling 12 month window for both longitudinal approach and cross-sectional approach. Choice of window a trade-off between “immediacy” and sample error

Contents | Prev | Next | Library

Desktop view | Switch to Mobile