Seite 63 - E_2012_01

Basic HTML-Version

63
01/2012
www.tradersonline-mag.com
TRADERS´
BASICS
F3)
Average Random Buy vs. Postponed Random Buy
Effect on the return of a trend following trade system when only entries
are taken when the Monest Value Indicator shows undervaluation.
Source: www.chartmill.com
(assuming it is a snake). So one
of our common ancestors already
must have developed this trait.
However, it takes far more than
a pattern to make a complete
trading system. The pattern on
which to enter a trade might well
be of less importance.
But if entries can be fine
tuned by adding our MVI as an
additional filter, it might certainly
be a good idea to put the idea
to the test of adding an MVI filter
to a complete trading system.
We backtested a really easy,
but totally objectively defined
trend following system that can
only enter a trade when the 25
bar simple moving average is
above the 75 bar simple moving
average. Trades are entered
when prices break above the
highest high of the previous five
bars. That is actually called a
five period Donchian Channel
breakout.
Again, we wanted to see what
happens, on average, with price,
up to 50 days after entry. The
result is shown in Figure 3 and
they are far from impressive. It
takes the average trade about 30
days to become only marginally
profitable.
Next we superimpose the trend
following trading system with
an MVI < -8 filter, meaning only
those five bar Donchian Channels
breakouts are taken when the
Monest Value Indicator has a
value below -8, a sign of short-
term temporary undervaluation.
Of course the moving averages
requirement also still holds.
The result here is quite
impressive. First, the average
trade has far less initial
drawdown, both in terms of
duration as well as in terms of
size. The maximum drawdown is
about half the original drawdown,
while the days the average trade
is in losing territory are minimised
to only about five to six days
(from almost 30 in the original,
un-enhanced, system). Secondly,
the average trade has an
overall much clearer trend. And
finally, compared 50 days upon
entry, the average trade for the
enhanced trend following trade
system has up to five times more
profit.
Conclusion
In our search towards a better
oscillator that produces sharper
and more objective signals
with the least lag, we built the
Monest Value Indicator based
on the concept of context.
Short-term valuation perception
being mainly lead by the most
recent prices, we used statistical
normalisation to capture an
objective interpretation of the
idea. However, the distribution
in bull and bear markets will be
skewed from perfectly normal,
meaning that under- and
overvaluation, now fixed at -8
and +8, could be calibrated onto
the real distribution. So, in a bull
market, undervalued probably will
have a slightly higher threshold
than -8. Likewise, in a bear
market, overvaluation perhaps
could be calibrated a little lower.
But as far as different financial
instrument were studied (futures,
commodities, equities, …) there
were no family specific, nor
product individual differences.
So a certain stock (of a certain
company) neither has a different
value distribution, nor a specific
one.
We conducted three back test
experiments. One experiment was
aimed at proving the standalone
quality of the Monest Value
Indicator in its own right. We
compared buying undervaluation
with buying at random, buying
overvaluation, buying on a dollar
cost averaging basis and a
combination of random entry with
undervaluation. An experiment
which made more than a nice
case for the quality of our new
breed of oscillator.
In a second and third
experiment we tried to answer
the question of whether the
Monest Value Indicator oscillator
could act as catalyst to enhance
both pattern performance and
system performances. And
though two experiments might
be too few to make a general
case, they seem very promising,
at least justifying further
research.