|
PureBytes Links
Trading Reference Links
|
There is a recent article in Stocks & Comms. on using any spreadsheet to
search for price patterns, the cup & handle formation is used as an
example. This uses correlation and could be set up with OLE from sheet to
MS to give trading signals. Have yet to try it. This method could also be
used to define indicator patterns; find the "ideal" one and trade on a
higher R2 for the same pattern for any security. Could also possibly use
correlation to some extent in MS. The article is in the Jan. 1998 issue,
page 62.
At 10:32 PM 2/4/98 -0600, you wrote:
>I am not quite sure how you would sum binary waves to obtain a binary
>number. Then again I might just be missing some of the math you are
>using.
>
>I was thinking of something along the lines of an ascending triangle.
>Lets say I want to identify the the following (in chronological order)
>S1 = lowest point for the triangle
>R1 = max of the first thrust which represents the resistence and the top
>side of the triangle
>S2 = low which follows R1 but higher than S1
>R2 = the max of the next thrust.
>
>My thoughts are that I can go through a couple of charts to get a range
>of values for % change from S1 to R1. Then use some value in % change
>from the low at S1 to high at R1. The goal here with % would be to
>screen out noise.
>
>I think the same thing would be true for R1 to S2. Screen with a strict
>inequality for S1 < S2.
>
>To find the general neighborhood of the points S1, R1, S2, R2 use the
>alert function.
>The idea would be to pair the points in alerts as follows:
>1: S1 < S2 in the alert over 5 to 10 days. (Looking at some charts
>will give some better numbers.)
>2: R1 > S1
>3: R1 > S2
>4: R2 > S2
>
>In the columns I would want to know the value of Abs( R1-R2).
>A possible representation of S1 and S2 would be: S1 := ref(llv(L,2),-3)
>and S2:= llv(L,2)
>
>Here I want a local neighborhood around S1 and S2 that are small. This
>is the LLV(L,2). I shift S1 back so the neighborhoods do not overlap.
>Hence S1: = ref(llv(L,2),-3).
>
>This is just the idea part. I hope you kind of get the idea of what I am
>thinking about.
>
>Harley
>
>Jim Michael wrote:
>
>> On Wed, 4 Feb 1998, Harley Meyer wrote:
>>
>> > This was the first time I have seen a post on pattern recognition
>> that
>> > could be used in Metastock. It would take some work to program for
>> > patterns. Also, by your example, I can see the extension into chaos
>> > theory.
>> >
>>
>> Something I have been thinking about to utilize MS for genetic
>> programming
>> is to use a sum of binary waves to produce an output which would be
>> equivalent to the pattern. If each binary wave is a power of ten, then
>> the
>> sum can be interpreted as a binary number (provided that wave n is
>> either
>> 10^n or 0), and converted to a symbol (e.g. hexadecimal
>> representation).
>>
>> Cheers,
>>
>> Jim
>> --
>> http://www.ilinks.net/~airesearch/marketalert/
>
>
>
>
>
RXR
|