Quick "theoretical" question for some of you who are more in the know: when computing the 'market' portfolio, it is supposed to be a market-cap weighted, right? What if your assets are non-equities? How can I utilize Black-Litterman with a mixed portfolio of stocks, bonds, and commodities? Is it possible to construct the 'market' portfolio in this case?
Thanks! I am really finding this fascinating to read about and try to implement.
EDIT: Does anybody have any available code for extended/augmented BL models? The basic model seems fairly trivial to implement (so long as you have the necessary optimization libraries), but some of the extended ones are throwing me for a loop. I am trying to work my way through Meucci's papers (in chronological order, when possible) and I would love to see an implementation of his some of the models he discusses in the papers from 2005-2010 (the latter, the better). Does anyone know if this is available, or is it all proprietary? I am really looking for extensions that deal with non-absolute performance metrics, e.g. 'Stocks ranked 1 will outperform stocks ranked 10 by an unknown amount with confidence 80%' or 'Energy stocks will have a negative return'. Thanks!
EDIT EDIT: Just stumbled across Copula-Opinion Pooling. Too cool...
Exploring Portfolio Optimization under Model Uncertainty
-
Corey
- Posts: 0
- Joined: Thu Jan 01, 2004 12:00 am
Exploring Portfolio Optimization under Model Uncertainty
"Then there was the man who drowned crossing a stream with an average depth of six inches." W. I. E. Gates
-
FatChoi
- Posts: 0
- Joined: Thu Jan 01, 2004 12:00 am
Exploring Portfolio Optimization under Model Uncertainty
Matlab examples from Meucci's book are available here.
-
Corey
- Posts: 0
- Joined: Thu Jan 01, 2004 12:00 am
Exploring Portfolio Optimization under Model Uncertainty
Anybody have any thoughts or know about any relevant discussions on the 'market' portfolio? Black Litterman says that if you don't have any views, you should be invested in the 'market' portfolio. However, my feelings are that a) the 'economists' market portfolio is not easily invested in, and b) the typical market-weighted portfolio is effectively momentum chasing, overweighting overvalued companies and underweighting undervalued companies.
My question is, has anyone had any success using Black Litterman, but instead of a market-weighted portfolio use a) an equal weighted portfolio, b) an earnings weighted portfolio, c) cash-flow weighted portfolio, or d) any other type of weighting system?
Thanks.
My question is, has anyone had any success using Black Litterman, but instead of a market-weighted portfolio use a) an equal weighted portfolio, b) an earnings weighted portfolio, c) cash-flow weighted portfolio, or d) any other type of weighting system?
Thanks.
"Then there was the man who drowned crossing a stream with an average depth of six inches." W. I. E. Gates
-
FatChoi
- Posts: 0
- Joined: Thu Jan 01, 2004 12:00 am
Exploring Portfolio Optimization under Model Uncertainty
The short answer is that the market portfolio is whatever your investors will accept as a benchmark.
The longer answer is that the theoretical CAPM market portfolio includes lots of untradeable assets -- individual future earnings for example -- that are not easily observable let alone investable. Even that answer generally involves making some pretty drastic rationalising assumptions. Fixed income can be very hard to fit into such a framework for example.
You might be right about the market portfolio although that's a relative verdict and doesn't address everything you need to worry about. The alternatives you list are trading strategies and not simply portfolios and so will present all kinds of practical difficulties. In any case if you pick one, you can calculate a set of implied market views that result in the other.
One thing to be careful of it the extent to which presentations of Black Litterman are single period while you may be interested in a continuous strategy. Mean reversion, a feature of the alternative portfolio weighting ideas above, may not look the same on a multiperiod basis.
The longer answer is that the theoretical CAPM market portfolio includes lots of untradeable assets -- individual future earnings for example -- that are not easily observable let alone investable. Even that answer generally involves making some pretty drastic rationalising assumptions. Fixed income can be very hard to fit into such a framework for example.
You might be right about the market portfolio although that's a relative verdict and doesn't address everything you need to worry about. The alternatives you list are trading strategies and not simply portfolios and so will present all kinds of practical difficulties. In any case if you pick one, you can calculate a set of implied market views that result in the other.
One thing to be careful of it the extent to which presentations of Black Litterman are single period while you may be interested in a continuous strategy. Mean reversion, a feature of the alternative portfolio weighting ideas above, may not look the same on a multiperiod basis.