In this thesis Genetic Progrmming is used to create trading systems for the EUR/USD foreign exchange market using intraday data. In addition to the exchange rates several moving averages are used as inputs.
The developed evolutionary algorithm extends the framework ECJ. The created trading systems are being evaluated by a fitness function that consists of a trading simulation. Genetic operators have been adapted to support "node weights". By using these on the one hand macromutaion is tried to be reduced on the other hand the interpretability of the created trading systems is tried to be improved.
Results of experiments show that created trading systems are apparently successfull in profitably using informations contained within the exchange rates. Profits of the created trading systems are maximized by using the optimal position size. It is shown that if the minimum investment period is met the achieved results are optimal even when taking into account the used risk adjusted performance figure.
Inhaltsverzeichnis (Table of Contents)
- Introduction
- Motivation
- Objective and structure
- Basic principles and state of the art
- Genetic Programming
- Program Structure
- Initialization of the GP Population
- The Genetic Operators
- Fitness Function
- Selection
- Process of the algorithm
- Crossover, building blocks and schemata
- Approaches against macromutation
- Modularization
- Further approaches for improvement
- Artificial Neural Networks
- Components of neural networks
- Network topologies
- Learning methods
- Trading Systems
- Tape Reader
- Market timing
- Position sizing
- Comparison of trading systems
- Fundamental versus technical analysis
- The Currency Market
- Approaches for the development of trading systems
- Genetic Programming
- Overview
- Requirements on the software
- Conception of software
- The Evolutionary Algorithm
- The fitness function
- Implementation
- Components of the developed software
- Classes of the exchange rate data server
- Classes of the Evolutionary Algorithm
- Overview over the framework ECJ
- Problems during experiments
- Experiment results
- Results with node weights
- Results of the training time period
- Results of the validation time period
- Results of the test time period
- Results as monthly turnovers
- Created trading rules
- Results without node weights
- Results of the training period
- Results of the validation time periods
- Results of the test period
- Results as monthly returns
- Created trading rules
- Identification and application of optimal f
- Results with node weights
- Discussion and evaluation
- Outlook
- Summary
Zielsetzung und Themenschwerpunkte (Objectives and Key Themes)
This thesis investigates the application of Genetic Programming (GP) in the development of trading systems for the financial market, particularly for the currency market. It explores how GP can be utilized to automate the process of finding effective trading strategies and examines the feasibility of generating trading systems that can outperform traditional methods. The key themes explored in this work include:- The application of Genetic Programming (GP) to financial markets.
- The development of trading systems based on GP-generated trading rules.
- The evaluation of the performance of GP-generated trading systems compared to traditional approaches.
- The exploration of the potential benefits and limitations of GP in the context of financial trading.
- The analysis of the factors influencing the effectiveness of GP-generated trading systems, such as market data quality and the design of the fitness function.
Zusammenfassung der Kapitel (Chapter Summaries)
Chapter 1: Introduction This chapter introduces the concept of Genetic Programming (GP) and its potential application in financial markets. It highlights the need for automated trading systems and the limitations of traditional approaches. The chapter also provides a background on the currency market and the challenges involved in developing effective trading strategies.
Chapter 2: Basic Principles and State of the Art This chapter provides a comprehensive overview of GP, including its core principles, program structure, initialization methods, genetic operators, fitness functions, selection techniques, and approaches to overcome limitations such as macromutation. It also delves into the concepts of artificial neural networks and their application in finance. The chapter concludes with a review of existing trading systems and approaches to their development.
Chapter 3: Overview This chapter introduces the overall approach and the requirements for the software developed in this thesis. It outlines the conception of the software, including the design of the Evolutionary Algorithm and the fitness function used to evaluate the performance of generated trading systems.
Chapter 4: Implementation This chapter describes the implementation details of the developed software. It covers the components of the software, the classes involved in data handling and exchange rate data server, the classes related to the Evolutionary Algorithm, and an overview of the ECJ framework used for GP implementation. Additionally, it discusses challenges encountered during the experimental phase.
Chapter 5: Experiment Results This chapter presents the results of experiments conducted using the developed software. It analyzes the performance of GP-generated trading systems with and without node weights, evaluating results based on various metrics such as training and validation periods, test periods, monthly turnovers, and monthly returns. The chapter also highlights the characteristics of the created trading rules.
Chapter 6: Discussion and Evaluation This chapter provides a comprehensive discussion of the obtained results. It analyzes the effectiveness of GP in generating trading systems, explores the limitations encountered during the study, and discusses the potential for future research.
Schlüsselwörter (Keywords)
The primary focus of this thesis lies in the development of trading systems using Genetic Programming (GP) for the currency market. The work explores the feasibility of automating the process of finding effective trading strategies through GP and examines the performance of GP-generated systems compared to traditional methods. This involves a comprehensive analysis of relevant concepts and techniques, including artificial neural networks, evolutionary algorithms, fitness functions, and the application of GP to financial markets. The key terms that encapsulate the core concepts and themes of this study are: Genetic Programming, Currency Market, Trading Systems, Evolutionary Algorithms, Artificial Neural Networks, Fitness Function, Automated Trading, and Financial Markets.- Quote paper
- Holger Hartmann (Author), 2007, Development of Trading Systems using Genetic Programming with a Case Study, Munich, GRIN Verlag, https://www.grin.com/document/186454