Are Sim Study Time Frames Limited By the Index Data?

[Eli]

Mitch--

I guess it makes sense, but I never thought of it until I actually
tried to run a sim study today. Even though I might want to model a
domestic equity market from 1970-1980...I cannot do so unless the
underlying index data correlated to the Mutual Fund Sim goes back that
far.

Furthermore...let's say I have built a portfolio in MCP to model a
clients portfolio containing 20 mutual funds. I have found the best
fit index to model each mutual fund...one mutual fund however is an
emerging market fund...the only index historical data I've found goes
back 8 years.

Because of the limitations of this one index in the overall portfolio,
I now can only create a sim study with a maximum number of 8 years....

Is this correct?

Thanks.

[Mitch]

That is correct. Emerging markets and other international markets
don't have deep index data available. This can create a number of
challenges for the Retirement Specialist. Here are my thoughts.

Asset Allocation is most important and the key to good asset
allocation is diversification. I would encourage you to think about
measuring the benefits using asset classes with long track records.
Without long track records, you will fall into the very problems that
you are trying to avoid: using incorrect data for building financial
models.

For me, I believe that asset classes should be broadly represented. I
would encourage an advisor to stick with the basic asset classes and
acquire broad representation. For example, if I wanted exposure to
domestic bonds, I would use the Lehman Bros. Aggregate Index. If I
wanted International exposure, I would use the MSCI EAFE index.

As for me asset allocation benefits can be achived by using the
following asset classes:
US Stocks
US Bonds
US Real Estate
International Stocks
Cash

Keep in mind that developed countries provide highly regulated markets
that eliminate corruption. Global companies domiciled in the US can
provide benefits of international market appreciation.

Keeping it simple can keep you from falling into the trap of becoming
an tactical allocator. And without historical data, there is no way of
measuring the potential benefits of including the asset class.

Mitch

[Eli]

I understand what you are saying and it makes sense. I agree that
this is the approach a Retirement Specialist should take when modeling
his/her own RECOMMENDED portfolios, but the real question pertains to
modeling the CURRENT portfolio of the potential client for future sim
study comparisons.

My understanding of the training to this point when modelling the
client's existing portfolio is to use MCP Stats to find the closest
correlating index to whatever Mutual Fund we are creating a sim for,
and yet...

Now I am wondering if it doesn't make better sense to find the closest
correlating index with the LONGEST historical data.

Without this longer historical data tied to the sim...it seems
difficult to really provide a quality sim study comparison between the
clients exisiting portfolio and the advisors recommended portfolio
over different time intervals

Your thoughts??

> > Thanks.- Hide quoted text -
>
> - Show quoted text -

[Mitch]

I believe that you will want to suffer the Correlation (R-Squared) for
the sake of historical data. The impact of historical data in the
modeling process is more important than having the highest
Correlation. However, don't simply accept a positive historical alpha
as the forecast.