We have previously illustrated that the monthly returns of companies such as Goldman Sachs
and
Morgan Stanley
are well explained by the
dynamic trading risk factor
. One interpretation of this analysis is to fault these companies for disguising their true nature,
but I feel that this is needlessly judgemental. The returns accruing to the risk factor one becomes exposed to by
trading in the markets have a non-negative mean and, apart from the drawdown per articulus, have performed quite
well over the past decade. It is surely wrong to fault these companies for pursuing a profitable line of business,
one just needs to be aware of where there returns truly originate (and it's not from advisory service or retail
brokerage).
Plenty of investors would like to be able to invest in a hedge fund and receive the returns they reap by their
trading activity. One can turn the analysis upside down and ask:
what companies should I own if I want the returns of a typical hedge fund?
We now have a simple answer to this question. If I want to invest in a hedge fund, but I am not sufficiently
wealthy or well connected, then I can invest in Goldman Sachs. This will get me exposure to the dynamic trading
risk factor without having to be an accredited investor, since GS is listed on the New York Stock Exchange.
However, let's take this analysis a little further since the R² for the Goldman Sachs regression
is around 50%, indicating other driving factors to their monthly returns. Suppose we took the five companies with
the largest R²s
from the S&P Select Sector Spider
for the financial services industry and built an equal weighted portfolio from them. How would this
portfolio's performance regress onto our factor series and how would such a portfolio perform? (n.b. The five
members is arbitary, but small portfolios are easy to manage.) One might call this A Poor Man's Hedge Fund,
since it should deliver the performance of a typical hedge fund without the requirement of dramatic wealth.
The chart above shows the performance of such a portfolio when the candidates are the current members
of the XLF. This portfolio's returns are very well explained by the dynamic trading risk factor, with an R²
of just under 70% (equivalent to a correlation coefficient of 83%). There is a selection bias associated
with examing the past performance of the current members of an index, for by definition the current membership
excludes any prior members that subsequently failed to meet the criteria for membership (or failed
completely), but in this case I do not believe this is an issue. We have tried to find the five members that are
most similar to the dynamic trading risk factor — not those that have performed the best. If our
results exclude a candidate which had a higher R², then a properly consituted series will have a
stronger effect. i.e. By not consituting the index properly (by not tracking the index membership through
time) we have, at worst, underestimated the tracking our our portfolio.