Do you have an easy way of remembering the subtle differencesin some of the 6 asset allocation models: if you do please share…thanks

Can u briefly state what these models are? That will help those of us who don’t have our books to know exactly what you’re talking about.

- Mean Variance 2. Rsampled Efficient Frontier 3. Black-Litterman 4. Monte Carlo Simulation (this one is fairly easy as the only multi period approach) 5. ALM (this one is easy) 6. Experienced Based (also ok) I guess I am trying to get a hold of the first 3…kind of confused of the details on the first three… Thanks

- Mean-Variance only uses one “run” based upon the estimated parameter inputs. Since the MV model is very sensitive to inputs, the efficient frontier can result in wacky outputs essentially. 2. This is where the re-sampled efficient frontier comes in. Instead of using one run or a point estimate, you can create a number of different efficient frontiers, and then average them to arrive at a re-sampled efficient frontier. 3. BL model extracts the mean returns, and variances from the market. So you get implied values. The BL model takes this one step further and states that you compare these extracted values to your own estimated valued and adjust according to your conviction. 4. Ditto - multi-period. Simulation based. Can take into effect different scenario’s, taxes, etc. It is a probability based model. 5. ALM - assets are allocated according to the nature and risk of the liability. 6. Rule of thumb heuristics.

ValueAddict Wrote: ------------------------------------------------------- > 1. Mean-Variance only uses one “run” based upon > the estimated parameter inputs. Since the MV > model is very sensitive to inputs, the efficient > frontier can result in wacky outputs essentially. > 2. This is where the re-sampled efficient > frontier comes in. Instead of using one run or a > point estimate, you can create a number of > different efficient frontiers, and then average > them to arrive at a re-sampled efficient > frontier. > 3. BL model extracts the mean returns, and > variances from the market. So you get implied > values. The BL model takes this one step further > and states that you compare these extracted values > to your own estimated valued and adjust according > to your conviction. > 4. Ditto - multi-period. Simulation based. Can > take into effect different scenario’s, taxes, etc. > It is a probability based model. > 5. ALM - assets are allocated according to the > nature and risk of the liability. > 6. Rule of thumb heuristics. If that is from memory VA, that is nothing less than spectacular! Respect.

- Mean Variance - one efficient frontier (once the frisbee shape curve moves to the right) which charts the highest return per unit of risk (highest sharpe ratio). - to avoid complexity of estimated returns, std dev’s and correlations, use historical data. 2. Resampled - multiple iterations of the mean variance frontier, in order to avoid concentrated weightings and undiversified portfolios 3. Black Litterman - Use a value weighted global market index to calculate implied equilibreum returns, which are then used in a mean variance approach. Correct me if I’m wrong here, but this is using the market’s “forward viewing” data to make forward estimates of returns, std dev’s, and correlations. 5. ALM - Remember, this is efficient ALM - so the highest surplus per unit of risk (not surplus return, surplus amount), so kind of like an information ratio in a sense.

from memory ha - don’t ask me swap questions tho - i will give you nothing in return

Don’t forget the sign-constrained mean-variance frontier

Thanks everyone…

Can someone confirm that the Black Litterman uses expecational data derived from the market to calc expected returns, std devs, correclation?

wow value addict, this is really helpful, thank you very much!

PhillyBanker: see this post on B-L portfolios http://www.analystforum.com/phorums/read.php?13,948546

I’m sitting here in a wedding reading af on my iphone, I think this makes me a bit OCD.