Creating portfolio risk and return models [36]

Go to: Summary | Previous | Next   
Bullet points include: Traditional time series analysis revolves around linear regression and least squares estimators If we believe we can forecast the future from the past then most obvious step is to include difference equations (c.f. GARCH) Behaviour of linear difference equations (e.g. p’th order) driven by roots of a p-th order polynomial, generating ARMA or ARIMA behaviour: Exponential growth or decay (or behaviour on the boundary, i.e. ‘integrated’) Regular (possibly damped or exploding) sinusoidal behaviour

Contents | Prev | Next | ERM Lecture Series

Desktop view | Switch to Mobile