Reverse Optimization

SS8-R17 Reverse Optimization

Trying to wrap my head around how exactly this works. My understand is that it takes the Weights drawn from MVO & adds in Covariance & Risk Aversion of the client, then spits out its own weights by making E.R.P the same and having unique betas for each asset class.

Basically it takes advantage by overweighing assets with higher returns relative to their beta, is that correct?

Also, how would we do we scale the asset classes so they all have similar ERP & unique betas ?

Because MVO is sensitive to minor changes input data, reverse optimization is sometimes preferred to obtain allocation weights. Asset weights in this approach are derived from the global market portfolio and used as input which “spit out” the output data being expected return, standard deviations and correlations. By using the global market portfolio as an anchor, the analyst’s ultimate allocation mix will be closer to the optimum and therefore improved. In the Black-Litterman approach the analyst adjusts the output from reverse optimization with his expectations of the market before he continues to retrieve asset weights.