|
PureBytes Links
Trading Reference Links
|
Re Dennis Holverstott:
You said:
"I think Schwager's point, and I agree, is that perpetual data
introduces
a bullish bias into the data. For example, the S&P futures will
currently have a premium of about 1200 points when they first become the
front month. This will decrease to near zero toward expiration.
Perpetual data removes this 400 points/month "downtrend". That may or
may not be a "lesser evil" for some particular system."
Aside from your example which I didn't have the time to analyze, the
CSI's Perpetual Contract calculation is designed to avoid any
positive(bullish) or negative(bearish) bias. It is strictly a
time-weighted calculation that focuses upon a constant forward period
from the current date measured in months and days that remains a
constant in the development of the resulting Perpetual series. The two
forward contracts that lie on either side of the forward period are
weighted in time so that the forward price will represent the time
weighted average of those two contracts.
If you want to see a case where there is a bullish (or bearish) bias
then you should consider a straight (non-proportional) back adjusted
contract and notice how often a longer term series actually goes
negative. The Perpetual approach will never go negative nor will the
proportional adjusted contract.
I never saw any of Schwager's writings that suggested he made the
claim that Perpetual Contract data had a bullish bias. Would you say
that London forward market's also have a forward bias? I don't think
so. The Perpetual Contract calculation is no different from the Forward
market method of presenting a time series. The two are essentially the
same. Schwager's position has always been that you can't trade the
actual Perpetual Data, but who said you could? Perpetual Contract data
represent a simple transformation that removes the expanding and
contracting birth-death process inherent in nearest future contract
information. It is an attempt to reconstruct the data into a form that
will at least remove some of the false alarm elements of the day-to-day
data that contributes to inevitable whipsaws in ones trading.
Analytically, it is just another tool that should not be taken as the
only element from which one makes a trading decision.
In any situation where a forward bias is present for whatever
reason, one should consider detrending the data to get the entire series
into today's dollar terms. The detrending will force those one cent
moves in corn back in 1949 to appear as significant as the 10 cent
moves of today. Or you might consider using a proportional adjustment
such as what has been advocated by Thomas Stridsman, an editorial staff
writer for Futures Magazine.
Like you said,
"The best the data vendors can do is provide a variety of methods for
building continuos files along with some educational guidelines about
which method may be best for various uses. The rest is up to the user of
the data. Buying a shiny new toolbox full of the best tools doesn't make
you a master carpenter. :-)"
There are many tools one should consider before placing your money on
the line.
And no master carpenter will throw out his tool box, and keep only his
hammer.
Regards,
Bob Pelletier
|