|
PureBytes Links
Trading Reference Links
|
At 7:08 AM -0400 6/29/00, editorial@xxxxxxxxxxxxx wrote:
>And so the debate rages on, year after year, system after system,
>without anyone going to the heart of the matter, to wit: Why does the
>system work in one time period and not in another? Why, gentlemen?
>The question is not whether, but WHY?
At 8:25 AM -0400 6/30/00, JPShark@xxxxxxx wrote:
>The reason systems work in some time periods and fail in others is
>because trading systems are simply "bandpass filters."
This explanation is a bit simplistic. This is one of my hot buttons
so pardon the "lecture" that follows.
Trading systems work because the person who designed it has found
some characteristic of the market that is correlated to what the
price will do in the future. This is fundamental. Without this, we
are just gambling.
If "X" occurs, then the price is likely to go up.
If "Y" occurs, then the price is likely to go down.
A trading system is programmed to look for X and Y and then take
appropriate action. Such characteristics are NEVER easy to spot. If
they were, every trader would be making money.
And the characteristic may only be apparent using some small range of
time compression. I know people who are trading patterns that appear
on 3 minute bars but the same pattern is not visible on 2 minute or 5
minute bars. Some things are apparent on tick-based charts that are
not apparent on time based charts, etc.
If one could simply pass the price through a bandpass filter and
trade the resulting signal, everyone would be rich since designing
bandpass filters is very easy and well known.
This is not to say that filters are not useful. If you are a
fisherman, it helps to wear Polaroid sunglasses to cut the glare to
help see the fish. But you still have to know where the fish are
likely to be hiding and what kind of disturbance to watch for to know
exactly where to catch fish. Few of the fisherman who buy Polaroid
sunglasses catch fish and those that do, don't often disclose their
REAL secrets.
Filters, such as Stochastics, RSI, MACD, etc., are like the Polaroid
sunglasses. They may cut the glare and make it easier to see the
"secret" pattern or characteristic you are looking for but used
alone, they are pretty much useless. A successful trader might tell
you that he just uses RSI, MACD, etc., and doesn't happen to mention
the "secret" characteristic that he is really looking for through
these "sunglasses".
I would go so far as to say that any characteristic that seems to be
present at any time compression is probably worthless for trading.
As to "why a system works in one time period and not in another",
that's easy. The "secret" pattern or characteristic you are trading
may come and go with time. Lots of markets change characteristics
during the lunch period, markets are different on Monday than on
Friday; in the spring and in the fall. Some characteristics disappear
when they become known by lots of people, other get reinforced.
Andy's system uses his Stochastic as a filter. Used alone, it is
probably worthless. But the combination of it with the Ehlers stuff
definitely works for the recent period. Maybe by studying it, we can
determine why it works and possibly make it work better, or make it
adaptive to different markets, etc.
Bob Fulks
|