|
PureBytes Links
Trading Reference Links
|
Dennis Holverstott wrote:
>I don't get it. In realtime, you don't have future bars available so >what
good is it for trading? You can make any filter *appear* to have >zero lag
by displacing the plot back in time.
>plot1[5](average(close,10), "zero lag");
>I'm sure your "zero lag" filter could be used to generate one hell of a
>good system report on a historical backtest but, unfortunately, you >can't
trade history. Please educate me about what I'm missing here.
>in case anyone missed my point, ****THIS SYSTEM IS A FAKE****. barry's
>dll COULD be used to produce fake results just like these. maybe he has
>some other use for it. i don't know. that's why i asked.
Peter Hoon wrote:
>The logic of what your are doing escapes me. Since any trading system you
>develop to trade real time, will not have any future date to access to
>develop low or near zero lag ma's, what is the point of developing dlls
or >software to eliminate this lag by accessing arrays of future historical
>data?
Apparently, some have not read all of my PowerBasic DLL postings in the
last month or so (including two source code examples), or have not pieced
the whole picture together. My purpose for doing a zero lag filter was
clearly stated as: first to take DLL functionality to a higher level and
second, I was curious to see how this type filter described by Tillson and
others would look. I made it very clear over and over again that it would
not have predictive value. I wasn't interested in it alone as a trading
tool (I did mention anchor for an anchored momentum indicator). Making
this filter work has shown that DLL's can be used for more important tasks
(use your imagination). One key is to run EL in multi-passes. In the
first pass, (from first to last bar in data series) all data you are
interested in is read into DLL arrays where it will then stay for whatever
you want to use it for on any future EL passes.
Keep your eye on the target. It ain't a particular indicator. It is
programming DLL's in BASIC rather than C+++++++, and extending and speeding
up EasyLanguage.
For what it's worth, this zero lag filter I posted does not look the same
as simply offsetting an average of the same type (T3, exponential, etc.) on
the chart. It looks much smoother and is a better fit. It can't be the
same as a single MA because it is the average of two MA's, one lagging the
data and the other leading the data.
|