Moving
Average Time Span
Moving Average can be constructed for any time period, whether a
few days, several weeks, many months, or even years. The
choice of length or time span is very important.
M
oving Average can be constructed for any time period, whether a
few days, several weeks, many months, or even years. The
choice of length is very important. For example, if it is
assumed that a complete bull and bear cycle lasts for 4
years, an Moving Average constructed over a time span longer than 48
months will not reflect the cycle at all. This is
because it smoothes out all the fluctuations that take
place during the period and will appear more or less as
a straight line crossing through the middle of the data.
On the other hand, a 5-day Moving Average will catch every minor
move in the stock cycle and will be useless for the
purpose of identifying the actual top and bottom of the
overall cycle.
Only an MA that can catch the
movement of the actual cycle will provide the optimum
trade-off between lateness and oversensitivity.
T
he choice of Moving Average depends on the type of market trend
that is to be identified: short, intermediate, or
primary. Because different markets have different
characteristics and the same markets go through
different cyclic phenomena, there is no such thing as a
perfect Moving Average. In recent years, extensive computer
research has been done on the optimum MA time span. The
conclusion from all sources is that there is no one
perfect time span.
What may work extremely
well in one market over one specific period of time is unlikely to be
duplicated in the future. When we talk about choice of
time span, we are really trying to identify an Moving Average that
will work most of the time with a specific time frame,
that is, short, intermediate, or long. Generally
speaking, long-term time spans are less influenced by
manipulation and knee-jerk random reactions to
unexpected news than are short term ones. This is why
long time spans usually give the best test results.
Examples of a 40-week span.
Research also shows that simple averages generally
outperform weighted and exponential ones.
T
he important thing to remember is that an MA is one
technical tool in the technical arsenal that is used
with other techniques as part of the art of
identifying trend reversals.