Daily Speculations
The Web Site of Victor Niederhoffer & Laurel Kenner
Dedicated to the scientific method, free markets, deflating ballyhoo,
creating value, and laughter;
| |
Daily Speculations The Web Site of Victor Niederhoffer & Laurel Kenner
Dedicated to the scientific method, free markets, deflating ballyhoo,
creating value, and laughter; |
Write to us at:
(not
clickable)
Q&A With Friends of Victor Niederhoffer: Alpha/Beta
4/21/2004
Q.
Over the years alpha and beta became the standard portfolio tools.
Then, as will happen, people started tweaking the standards to try
and get a bit more out. Amongst the tweaks are adjusting alpha for
volatility and adjusting relative strength (a composite of alpha
and beta) for volatility. For example, one practitioner I know uses
an alpha/sd criteria for fund selection and another uses beta
adjusted rs ranks for stock selection. Both are very happy with
their methods.
I remain very uncomfortable with these ideas. Do any Specs have any
thoughts on adjusting relative returns and/or unique returns by
volatility--pro or con?
A.
Philip J.
McDonnell: I have mixed thoughts on this subject. On the one hand, any measure of
return should include some estimate of how much risk was assumed to achieve
same. Thus some measure which includes volatility or beta should be part of
any investment analysis.
However most such measures use the risk measure (beta/volatility) as a
divisor and are susceptible to anomalous behavior as risk approaches zero.
For example what is the Sharpe ratio of a riskless t-bill? The definition
says it is the excess return divided by the volatility. Is it zero because
by definition it cannot exceed its own return? Or is it infinite because we
are dividing by zero the presumed riskless divisor?
The above may seem to be an arcane academic discussion to some but if we
consider the market neutral hedge funds which have very low volatility and
very low returns. As a class they tend to dominate the rankings of hedge
funds when sorted by Sharpe ratio. But looking at their absolute returns
they are typically on the order of 2 to 3% per year. The anomaly is caused
by the aforementioned issue of a near zero volatility divisor.
In my opinion the best measure of performance includes absolute return and
separately considers some measure of risk as an absolute number. It is
appropriate to include some sort of ratio measure but by itself such a
measure is not enough.
A.
Kurt Specht: Here's one related to mutual fund selection:
"In 1998, the author developed a more aggressive volatility-adjusted
performance measure called NCAlpha, which reduces the achieved daily
returns of the fund by the returns of the relevant market index,
amplified by the relative volatility of the fund to the market index.
If a fund s daily returns are twice as high as the index and its daily
volatility is twice has high as the index, its volatility-adjusted
return is zero. Since NCAlpha is based on relative volatility and is
tied to a market index, high NCAlpha-ranked funds may be more volatile
but achieve higher returns that those for the Sharpe Ratio."
http://www.madriver.com/~wwgansz/Volatility_Adjusted_Mutual_Fund_Selection.htm
From what I have seen though, it has not panned out better than fund
selection via Sharpe.
Readers may send questions on finance and investment to
us c/o 
(address
is not clickable). If we don't know the answer, we'll
submit it our experts.