|
PureBytes Links
Trading Reference Links
|
At 11:56 AM -0500 1/17/99, Ira wrote:
>One uses moving averages to indicate whether the item being traded is
>rising or falling in price. If that is the assumption, then logic and
>several very interesting books put forth the thesis that the two moving
>averages should be the span and 1/2 span of the cycle being traded. For a
>true reading the moving averages should be offset by 1/2 their cycle
>length in order to reflect the moving average of the bar above it
>accurately.
There is a good mathematical reason for this. The simple moving average has
no response to a price cycle equal to the length of the moving average. You
can see this simply by imagining a price cycle that repeats exactly every
20 bars and a moving average of 20 bars in length. Thus on a new bar, the
price value that drops off the end of the average is exactly equal to the
value that you add for the new bar so the value of the moving average stays
exactly the same. But the response of a 10 bar moving average will be
significant. (An engineer would refer to the frequency response of the
moving average as having a "null" at a period of 20 bars.)
>Part of the accuracy in some of the moving averages is self fulfilling. If
>everyone believes that the 200 day moving average or the 50 week moving
>average is the holy grail, then when they are penetrated the buy or sell
>programs will follow as the traders act like sheep. In commodities the 18
>and 9 bar averages seem to be the popular ones. In stochastic everyone
>seems to use 14 as the norm. Because it is popular does not make it right.
The significance of crossing the popular 50 day or 200 day average line is
totally psychological and has nothing to do with anything fundamental. If
enough people think the price will keep dropping once it penetrates the 50
day average, it will because they will all sell, forcing the price down.
There are trading strategies that capitalize on this phenomenon.
Bob Fulks
|