Hitoshi Iba and Claus C. Aranha Practical Applications of Evolutionary Computation to Financial Engineering Robust Techniques for Forecasting, Trading and Hedging 4Q Springer
Contents 1 Introduction to Genetic Algorithms 1 1.1 What Is Evolutionary Computation? 2 1.2 Basic Principles of Evolutionary Computation 6 1.2.1 Introduction to GA 6 1.2.2 Introduction to GP 10 1.3 WhyGAandGP? 15 2 Advanced Topics in Evolutionary Computation 19 2.1 Multi-objective Optimization :. 20 2.1.1 Risk or Return?.\ 20 2.2 Memes and Memetic Algorithms 25 2.2.1 Memes - The Cultural Genes 25 2.2.2 Can the Meme Survive in the World of Finance? 27 2.2.3 The Memetic Algorithm 28 2.2.4 Baldwinian Evolution 30. 2.2.5 Baldwin Effects on FX Trading Rule Optimization 34' 2.3 Real-Valued GAs 38 2.3.1 Differential Evolution 38 2.3.2 Particle Swarm Optimization 45 2.4 Randomness Issues in Evolutionary Computation 51 2.4.1 Tree Generation for Genetic Programming 52 2.4.2 Experiments with Predicting Time-Series Data 55 3 Financial Engineering 61 3.1 Basic Concepts in Financial Engineering 62 3.1.1 The Technical and Fundamental Approaches 62 3.1.2 Market Elements, 63 3.1.3 Technical Trading Concepts 63 3.2 Price Prediction 64 3.2.1 Option Pricing and the Black-Scholes Formula 64 3.3 Trend Analysis 69
X Contents 3.3.1 Strategies for Trend Analysis 70 3.3.2 Overview of Technical Indicators 70 3.4 Automated Stock Trading 76 3.5 Portfolio Optimization 77 3.5.1 Problem Definition 78 3.5.2 Evolutionary Approaches to Portfolio Optimization 82 4 Predicting Financial Data 85 4.1 Methods for Time Series Prediction 85 4.2 STROGANOFF 89 4.2.1 GMDH Process in STROGANOFF 89 4.2.2 Crossover in STROGANOFF....7.' 94 4.2.3 Mutation in STROGANOFF 95 4.2.4 Fitness Evaluation in STROGANOFF 95 4.2.5 Recombination Guidance in STROGANOFF 95 4.2.6 STROGANOFF Algorithm 97 4.3 Application to Financial Prediction 98 4.3.1 STROGANOFF Parameters and Experimental Conditions 98 4.3.2 GP Parameters and Experimental Conditions 99 4.3.3 Validation Method 99 4.3.4 Experimental Results 101 4.3.5 Comparative Experiment with Neural Networks 105 4.4 Inductive Genetic Programming 106 4.4.1 Polynomial Neural Networks 107 4.4.2 PNN Approaches 107 4.4.3 Basic IGP Framework Ill 4.4.4 PNN vs. Linear ARMA Models 112 4.4.5 PNN vs. Neural Network Models 114 4.4.6 PNN for Forecasting Cross-Currency Exchange Rates... 116 4.5 Challenging Black-Scholes Formula 119 4.6 Is the Precise Prediction Really Important? 120 5 Trend Analysis 123 5.1 The Data Classification Problem 124 5.2 The MVGPC * 127 5.2.1 Classification by Genetic Programming 127 5.2.2 Majority Voting System 133 5.3 Applying MVGPC to Trend Analysis 137 5.4 MVGPC Extension 139 6 Trading Rule Generation for Foreign Exchange (FX) 141 6.1 Automated Trading Methods Using Evolutionary Computation... 142 6.1.1 Applications of GA and GP 142 6.1.2 Application of PSO and DE 144
Contents XI 6.2 Price Prediction Based Trading System 145 6.2.1 Trend Prediction 145 6.2.2 Generating Trading Rules 146 6.2.3 Experimental Results 148 6.3 The GA-GP Trading System 149 6.3.1 Why Optimize Indicators' Parameters? 149 6.3.2 Fitness Function 150 6.3.3 Implementation of the GA-GP System 152 6.3.4 Practical Test of the GA-GP System 158 6.4 Using DE and PSO for FX Trading 162 6.4.1 Moving Average FeaturevBased Trading System 163 6.4.2 Dealing Simulation 168 7 Portfolio Optimization 175 7.1 A Simple GA for Portfolio Optimization 178 7.1.1 Genome Representation 179 7.1.2 Evolutionary Operators 180 7.1.3 Selection Method 181 7.1.4 Fitness Function 183 7.1.5 Testing the Array-Based GA 184 7.2 MTGA - GA Representation for Portfolio Optimization 187 7.2.1 Main Strategies of the MTGA 188 7.2.2 Implementation of the MTGA,189 7.2.3 Hybridization Policy 194 7.2.4 Test-Driving the MTGA 195 7.3 Implementation Issues for Portfolio Optimization 198 7.3.1 Dynamic Data and Portfolio Rebalancing 198 7.3.2 Asset Lots and Portfolio Weighting 199 7.3.3 Trader Policies 200 7.3.4 Alternative Risk and Return Measures 201 A Software Packages 203 A.I Introduction." 203 A.2 Multi-objective Optimization by GA ''... 203 A.3 Time Series Prediction by GP 207 A.4 Majority Voting GP Classification System 208 A.5 STROGANOFF Time Series Prediction and System Identification. 209 A.6 Portfolio Optimization Testing Suite 214 B GAGPTrader 219 B.I System Requirements 219 B.2 Preparing the GAGPTrader for Installation 220 B.2.1 Meta Trader 4 Demo Account 223 B.3 GAGPTrader Trial Version Installation 224
XII Contents B.4 Setting Up the GAGPTrader Trial Version 227 B.5 Parameters 230 B.6 Output Log 232 B.7 Description of the Symbols Used in the Charts 233 References 235 Index 243