Mining financial time-series using symbolic analysis

Pao Yue-kong Library Electronic Theses Database

Mining financial time-series using symbolic analysis

 

Author: Chau, Man-chuen
Title: Mining financial time-series using symbolic analysis
Degree: M.Sc.
Year: 2008
Subject: Hong Kong Polytechnic University -- Dissertations.
Time-series analysis -- Data processing.
Data mining.
Econometrics.
Department: Dept. of Computing
Pages: ix, 71 leaves : ill. ; 30 cm.
Language: English
InnoPac Record: http://library.polyu.edu.hk/record=b2231905
URI: http://theses.lib.polyu.edu.hk/handle/200/1439
Abstract: There are many financial and mathematical tools for analyzing and describing financial time-series data. Dow Theory, Bar Charts and Candlestick Charts are very popular and a good presentation of stock's momentum. These technical analyses deem that buyers and sellers move markets based on their individual expectation and their individual emotion. All what they have known will reflect in the market price. Due to market inefficiency (no perfect market), the market price of a financial product may not reflect the underlying value. However, those tools do not treat the data as streaming and cannot let Financial Analysts know one important piece of information regarding a stock's momentum; they cannot incorporate the analysis with transaction volume into one chart but Financial Analysts are often more concerned about how big the trading volume (quantity) is, especially when the price has turned the direction up or down. The characteristics of financial time-series data are discrete and streaming though data stream must stop during non-trading period. The opening price of the current trading session cannot continuously link with the closing price of the previous trading session in most cases. Hence, price gap may exist between two trading sessions. Financial Analysts are often concerned about when this price gap will be filled up thereafter. In this work, financial time-series data is transformed into different symbolic representations so as to perform the relevant symbolic analysis and data mining for time-series forecasting. Data may be divided into many equal intervals of time session (i.e. half day, one day or one week etc). A pre-defined symbolic object, according to its opening value, closing value, highest value, lowest value and the trading volume, can represent a pattern of each time session. Trading volume will be normalized because the primary analysis aims at how big the volume is, not the exact value of the volume. Two or more consecutive time sessions may also be retransformed together into a new symbol for dimensionality reduction. Every symbolic object is used to plot a chart for visual analysis. Financial analyst can find the patterns of price trend, trading volume and price gap on the chart. The raw data will be processed and transformed into many symbolic objects. The analysis will also output a series of symbols, which when fed to the decision tree approach of classification to the symbols, will contribute to predicting subsequent trading pattern(s). The final result may help Financial Analysts in the trading strategy.

Files in this item

Files Size Format
b22319050.pdf 6.874Mb PDF
Copyright Undertaking
As a bona fide Library user, I declare that:
  1. I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
  2. I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
  3. I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.

     

Quick Search

Browse

More Information