/


Market Risk [19]

Go to: Summary | Previous | Next   
Bullet points include: Assume losses follow: Estimate volatility using weighted average of lagged squared, de-meaned returns. Typically weights constructed with exponential decay, i.e. lambda i proportional to lambda ^ i for some constant lambda (with weights then scaled to sum to unity). See also www.nematrian.com/ExtremeEventsQuestionsAndAnswers2_3q.aspx . Issues include: how wide should data window be, what should the weights (decay factor) be, should mean returns be estimated or set to zero?

NAVIGATION LINKS
Contents | Prev | Next | ERM Lecture Series


Desktop view | Switch to Mobile