|
PureBytes Links
Trading Reference Links
|
The old zig-zag catches many in its trap. Another way to distort findings
are to buy/sell at open with no delay. You buy before the signal occurs. If
you buy/sell at close, with no delay, you are required to know the signal
occurred with out having computed it. These types of distortions make many
a system look good. I won't even get into optimization and what it can do.
The choice of markets and stocks for test are also a big help in blinding
the user to true results. I think you need to run tests on a min of 500
stocks with all sorts of life cycles present. Ranging, trending, long term
decline as well as going up long term. That is why MSWIN's one stock at a
time is ridiculous.
Please note also that when you have an indicator with a length of 250
periods, you have delayed first trade by a minimum of one year. While you
wait for the year to past, buy and hold jumps ahead or falls back depending
on the length of your database and the year it started. The person with a
short database or a stock just entering the market in last 2 years or so
finds a lack of data to determine goodness or badness of system.
Richard Estes
|