Detecting Regime Change in Computational Finance: Data Science, Machine Learning and Algorithmic Trading

Length: 138 pages
Edition: 1
Language: English
Publisher: Chapman and Hall/CRC
Publication Date: 2020-09-15
ISBN-10: 0367536285
ISBN-13: 9780367536282
Sales Rank: #2703072 (See Top 100 Books)

Based on interdisciplinary research into “Directional Change”, a new data-driven approach to financial data analysis, Detecting Regime Change in Computational Finance: Data Science, Machine Learning and Algorithmic Trading applies machine learning to financial market monitoring and algorithmic trading. Directional Change is a new way of summarising price changes in the market. Instead of sampling prices at fixed intervals (such as daily closing in time series), it samples prices when the market changes direction (“zigzags”). By sampling data in a different way, this book lays out concepts which enable the extraction of information that other market participants may not be able to see. The book includes a Foreword by Richard Olsen and explores the following topics:

Data science: as an alternative to time series, price movements in a market can be summarised as directional changes
Machine learning for regime change detection: historical regime changes in a market can be discovered by a Hidden Markov Model
Regime characterisation: normal and abnormal regimes in historical data can be characterised using indicators defined under Directional Change
Market Monitoring: by using historical characteristics of normal and abnormal regimes, one can monitor the market to detect whether the market regime has changed
Algorithmic trading: regime tracking information can help us to design trading algorithms

It will be of great interest to researchers in computational finance, machine learning and data science.

About the Authors

Jun Chen received his PhD in computational finance from the Centre for Computational Finance and Economic Agents, University of Essex in 2019.

Edward P K Tsang is an Emeritus Professor at the University of Essex, where he co-founded the Centre for Computational Finance and Economic Agents in 2002.

Cover
Half Title
Title Page
Copyright Page
Table of Contents
Foreword
Preface
List of Figures
List of Tables
Chapter 1 Introduction
    1.1 Overview
    1.2 Research Objectives
    1.3 Book Structure
Chapter 2 Background and Literature Survey
    2.1 Regime Change
        2.1.1 Regime Change Detection Methods
    2.2 Directional Change
        2.2.1 The Concept of Directional Change
        2.2.2 Research Using Directional Change
        2.2.3 Directional Change Indicators
            2.2.3.1 Total Price Movement
            2.2.3.2 Time for Completion of a Trend
            2.2.3.3 Time–Adjusted Return of DC
    2.3 Machine Learning Techniques
        2.3.1 Hidden Markov Model
            2.3.1.1 Definition of HMM
            2.3.1.2 Parameters of HMM
            2.3.1.3 Expectation-Maximization Algorithm
        2.3.2 Naïve Bayes Classifier
            2.3.2.1 Definition of Naïve Bayes Classifier
Chapter 3 Regime Change Detection Using Directional Change Indicators
    3.1 Introduction
    3.2 Methodology
        3.2.1 DC Indicator
        3.2.2 Time Series Indicator
    3.3 Experiments
        3.3.1 Data Sets
        3.3.2 Hidden Markov Model
    3.4 Empirical Results
        3.4.1 EUR–GBP
        3.4.2 GBP–USD
        3.4.3 EUR–USD
        3.4.4 Distribution of the Indicator R
        3.4.5 Discussion
    3.5 Conclusion
Chapter 4 Classification of Normal and Abnormal Regimes in Financial Markets
    4.1 Introduction
    4.2 Methodology
        4.2.1 Summarising Financial Data in DC
        4.2.2 Detecting Regime Changes through HMM
        4.2.3 Comparing Market Regimes in an Indicator Space
    4.3 Empirical Study
        4.3.1 Data Sets
        4.3.2 Summarising Data under DC
        4.3.3 Detecting Regime Changes under HMM
        4.3.4 Observing Market Regimes in the Normalised Indicator Space
    4.4 Results and Discussions
        4.4.1 Market Regimes in the Indicator Space
        4.4.2 Market Regimes under Different Thresholds
        4.4.3 Discussion
    4.5 Conclusions
Chapter 5 Tracking Regime Changes Using Directional Change Indicators
    5.1 Introduction
    5.2 Methodology
        5.2.1 Tracking DC Trends
        5.2.2 Use of a Naïve Bayes Classifier
    5.3 Experiment Setup
        5.3.1 Data
        5.3.2 Regime Changes on the Data
    5.4 Empirical Results
        5.4.1 Calculating Probability
        5.4.2 B-Simple for Regime Classification
        5.4.3 B-Strict for Regime Classification
        5.4.4 Tracked Regime Changes
            5.4.4.1 Tracked Regime Changes on DJIA Index
            5.4.4.2 Tracked Regime Changes on FTSE 100 Index
            5.4.4.3 Tracked Regime Changes on S&P 500
        5.4.5 Discussion
    5.5 Conclusion
Chapter 6 Algorithmic Trading Based on Regime Change Tracking
    6.1 Overview
    6.2 Methodology
        6.2.1 Regime Tracking Information
        6.2.2 Trading Algorithm JC1
        6.2.3 Trading Algorithm JC2
        6.2.4 Control Algorithm CT1
    6.3 Experimental Setup
        6.3.1 Data
        6.3.2 Experimental Parameters
        6.3.3 Money Management
    6.4 Experiment Results
        6.4.1 Number of Trades
        6.4.2 Final Wealth
        6.4.3 Maximum Drawdown
    6.5 Discussions
        6.5.1 The Primary Goals Are Achieved
        6.5.2 Future Work: Regime Tracking for Better Trading Algorithms
    6.6 Conclusions
Chapter 7 Conclusions
    7.1 Summary of Work Done
    7.2 Take-Home Messages
    7.3 Future Research
        7.3.1 Research Directions
Appendices
    Appendix A A Formal Definition of Directional Change
    Appendix B Extended Results of Chapter 3
    Appendix C Experiment Summary of Chapter 4
    Appendix D Detected Regime Changes in Chapter 4
Bibliography
Index