Serving the Quantitative Finance Community

 
JohnFrusciante
Topic Author
Posts: 5
Joined: January 21st, 2019, 10:50 am

RiskReturn Optimization using an 'ad hoc' risk measure

January 31st, 2019, 5:46 pm

Hi

within the end year AA for the company I work for we are performing some risk-returns optimization based on different risk measures.
Given a set of simulated returns, the idea is to build up a measure which averages the squares of the difference between some percentiles of the portfolio return distribution (at a certain projection step) and the expected portfolio return.

So here's the approach: 
1) I'd generate a big (enough) set of random weights to compute the expected return of the portfolio to be used within the risk function to be minimized

2) I'd build an objective function (i.e. this risk function) to be minimized by a set of optimal portfolio weights. In order to do this, I would create a matlab function handle where the percentiles of the portfolio return distribution are expressed as a function of ptf weights (and calculate the difference wrt the expected return calculated in 1).

I am not sure whether this is consistent. Could someone please help me out? thanks in advance!