Information Technology Reference
In-Depth Information
Price Trackers Inspired by Immune Memory
William O. Wilson, Phil Birkin, and Uwe Aickelin
School of Computer Science, University of Nottingham,UK
{
wow, pab, uxa
}
@cs.nott.ac.uk
Abstract.
In this paper we outline initial concepts for an immune in-
spired algorithm to evaluate price time series data. The proposed solu-
tion evolves a short term pool of trackers dynamically through a process
of proliferation and mutation, with each member attempting to map
to trends in price movements. Successful trackers feed into a long term
memory pool that can generalise across repeating trend patterns. Tests
are performed to examine the algorithm's ability to successfully identify
trends in a small data set. The influence of the long term memory pool
is then examined. We find the algorithm is able to identify price trends
presented successfully and e
ciently.
1
Introduction
The investigation of time series data for analysis and prediction of future in-
formation is a popular and well studied area of research. Historically statistical
techniques have been applied to this problem domain, however in recent years
the use of evolutionary techniques has seen significant growth in this area. Neu-
ral networks [6] [13], genetic programming [7], and genetic algorithms [3] are all
examples of methods that have been recently applied to time series evaluation
and prediction.
However the use of immune inspired (IS) techniques in this field has remained
fairly limited [9]. IS algorithms have been used with success in other fields such
as pattern recognition [2], optimisation [5], and data mining [8]. In this paper we
propose an IS approach, using
trackers
to identify trends in time series data, and
take advantage of the associative learning properties exhibited by the natural
immune system.
The time series proposed for investigation in this paper is that of price move-
ments (Section 2) and the approach used to identify trends in price data is
inspired by the immune memory theory of Dr Eric Bell [1]. His theory indicates
the existence of two separately identifiable memory populations which are ide-
ally suited to recognise long and short term trends prevalent in time series data.
In Section 3 we discuss this immune memory theory and introduce other im-
mune mechanisms which form part of our algorithm. The algorithm itself is then
presented in Section 4. The methodology for testing the algorithm, the results
and discussions of the results are documented in Sections 5, 6 and 7 respectively,
before concluding in Section 8.
