|
PureBytes Links
Trading Reference Links
|
First, thanks to all who replied.
Scott- Just to clarify: You optimize your portfolio returns based on
some return/risk metric. Then you examine the correlation matrix. Then
you throw out spurious correlations based on logic and market intuition.
Do you do this by strategy or by commodity? That is, do you optimize by
the expected returns of a commodity over all systems or by the expected
returns of systems themselves with some standardized weights? Lastly,
could you clarify your comment about making no assumptions about future
returns?
Kevin- My understanding is that the turtles rules just assigned
percentage weights based on dollar volatility and kept total exposure of
correlated groups to some minimum level. Is that right?
VK-To test portfolios in TS you have to buy some 3rd party software. I
bought MSC Pro and have not used it enough for a evaluation.
Thanks again,
Gabriel
-----Original Message-----
From: Scott Hoffman [mailto:trader2020@xxxxxxx]
Sent: Tuesday, January 27, 2004 6:24 PM
To: omega-list
Subject: Re: Portfolio Construction
I focus a lot of my attention to portfolio construction. My input would
be:
1) Get familiar with the concepts of Modern Portfolio Theory.
2) Use the correlation of market return streams, not raw price data.
3) Get an MVO (Mean-Variance Optimizer) tool and play around with it.
4) Don't assume future correlations unless they make sense, ie.
CL/HO/HU, or US/TY.
5) Don't make any assumptions about future returns from any one market,
ie the worst market in the past may be the best one going forward.
The more uncorrelated (negatively correlated is the best, but hard to
find) market return streams you can combine, the higher the
risk-adjusted return (like Sharpe ratio) of the portfolio will be.
Scott Hoffman
Red Rock Capital Management, Inc.
----- Original Message -----
From: "Gray, Gabriel" <Gabriel.Gray@xxxxxxxxxxxxxxxxx>
To: <omega-list@xxxxxxxxxx>
Sent: Tuesday, January 27, 2004 12:24 PM
Subject: Portfolio Construction
> Hi All,
>
> Has anyone done any work with, or have a strong opinion about,
> correlations between commodities and/or correlations between
> strategies. Has anyone read anything, measured the effect of, or
> optimized portfolios so as to maximize negative correlation or
> construct the most efficient portfolio. I am obviously very new to
> this so any advice would be much appreciated.
>
> Gabriel
>
>
|