Creating portfolio risk and return models [45]

Go to: Summary | Previous | Next   
Bullet points include: Tries to tackle problem of sensitivity to the inputs E.g. possible approaches to tackling estimation error Avoid relying exclusively on past data (and include your own or market views?) User higher frequency data (but how then to knit this data together?) Impose factor structure (i.e. your own prior views?) Use shrinkage approach (but doesn’t seem to change portfolio mix?) Impose constraints (e.g. short-sale, diversification, 1/N) Bayesian approaches, e.g. Black-Litterman Kemp (2010) also reviews resampled optimisation, but it too is not a panacea

Contents | Prev | Next | ERM Lecture Series

Desktop view | Switch to Mobile