Re: Pattern Recognition in Stock Market Data

• To: mathgroup at smc.vnet.net
• Subject: [mg22364] Re: Pattern Recognition in Stock Market Data
• From: Keith Erskine <keithe69 at ozemail.com.au>
• Date: Fri, 25 Feb 2000 21:14:22 -0500 (EST)
• Sender: owner-wri-mathgroup at wolfram.com

```Hello:

I wish to get advice/thoughts/ideas as to whether Mathematica can provide a
solution to the following.  It relates to scanning a set of files of
price-time data of stockmarket data (eg the Australian Stock Exchange -
ASX).  This is end-of-day data, which has Open-High-Low-Close-Volume for
each stock traded on the day.  We are talking about 3000 separate files to
search through.

that may lead to what is termed a 'breakout' in price.  These are movements
that can be quite profitable.

I wish to scan through this end-of-day data, looking for what can be termed
an emerging pattern, that is, BEFORE the breakout occurs.  Where this
emerging pattern is detected, these particular stocks are put into a 'watch
list'.  Some will breakout, others will not, there are no guarantees. But
those in the watch list can be monitored, and a breakout then detected by
other means.

One such pattern that occurs all the time is the ascending triangle.  See
simplistic figure below:
^
|
X
X Price Breakout
X x x x x x x x x x x x Resistance
X x x x x x x x x x
X x x x x x x x
X x x x x x
X x x x

Time --->

Above is the most simplistic representation of all, where each column of
x's represents the High-Low price range for each successive day.  There can
be quite abstract variations on this pattern, with oscillatory type
behaviour within the triangle, but nevertheless the price sooner or later
converges towards the top side of the ascending triangle, known as a line
of price resistance.  Once this resistance is broken, typically the price
can keep going, hence the term breakout.

Any suggestions as to whether this problem is suited to Mathematica, or can
provide a guide to how this would be solved using Mathematica would be
greatly appreciated.