|
Title:
|
A DECISION TREE APPROACH FOR SECURITY A DECISION TREE APPROACH FOR SECURITY |
|
Author(s):
|
Gilles Richard , Hasan Alsaedy |
|
ISBN:
|
978-972-8924-86-7 |
|
Editors:
|
Hans Weghorn, Jörg Roth and Pedro Isaías |
|
Year:
|
2009 |
|
Edition:
|
Single |
|
Keywords:
|
Trading, decision tree, prediction, accuracy rate |
|
Type:
|
Full Paper |
|
First Page:
|
27 |
|
Last Page:
|
33 |
|
Language:
|
English |
|
Cover:
|
|
|
Full Contents:
|
click to dowload 
|
|
Paper Abstract:
|
In this paper, we investigate the relevance of a classical machine learning technique as a tool for securities trading
technical analysis. Our main idea is to build up, starting from the data at hand, a decision tree which can be used as a
predictive model. Getting financial data, we investigate the past history of a given security, and we code this history as a
word over the binary alphabet {u (for up), d (for down)}, keeping only the final output of a trading day. Starting from the
empirical fact that only 5 days of trading history are sufficient to guess the current day, we breakdown our history word
into examples of 6 trading days, the last one being the one to be guessed. Doing so, we get a robust training set with
which we can run a classical decision tree making algorithm. After relevant tuning of our initial methods and tests over
diverse securities freely available on the Internet it appears that our method provides very interesting accuracy rate and
gives hope for the future. It appears as well that an increase of the past trading window (from 5 to 10 days) does not
significantly improve the accuracy rate. |
|
|
|
|
|
|
