Artificial Intelligence in Financial Markets (eBook)

Dr Christian L. Dunis is a Founding Partner of Acanto Research, where he is responsible for global risk and new products. He is also Emeritus Professor of Banking and Finance at Liverpool John Moores University where he directed the Centre for International Banking, Economics and Finance (CIBEF) from February 1999 through to August 2011. Christian holds a MSc and a Superior Studies Diploma in International Economics, and a PhD in Economics from the University of Paris. Dr Peter W. Middleton completed his PhD at the University of Liverpool. His working experience is in Asset Management and he has published numerous articles on Financial Forecasting of Commodity spreads and Equity time series. Dr Andreas Karathanasopoulos studied for his MSc and Phd at Liverpool John Moores University under the supervision of Professor Christian Dunis. His working experience is academic having taught at Ulster University, London Metropolitan University and the University of East London. He is currently Associate Professor at the American University of Beirut and has published over 30 articles and one book in the area of artificial intelligence. Dr Konstantinos Theofilatos completed his MSc and PhD in the University of Patras Greece. His research interests include computational intelligence, financial time series forecasting and trading, bioinformatics, data mining and web technologies. He has published 27 publications in scientific peer reviewed journals and over 30 articles in conference proceedings.

Preface 5
Contents 6
The Editors 10
Acknowledgements 10
Final Words 10
References 11
Contents 13
Notes on Contributors 15
Part I: Introduction to Artificial Intelligence 16
1: A Review of Artificially Intelligent Applications in the Financial Domain 17
1 Introduction 17
Applications of ANN in Finance 22
Portfolio Management 23
Stock Market Prediction 24
Risk Management 25
2 Application of Expert Systems in Finance 25
Portfolio Management 26
Stock Market Prediction 33
Risk Management 33
3 Applications of Hybrid Intelligence in Finance 34
Portfolio Management 34
Stock Market Prediction 37
Risk Management 38
4 Conclusion 42
5 Appendix 1 43
Regression Analysis [7] 43
Classification [7] 44
Clustering [7] 45
Fuzzy c-means clustering [7] 46
Back propagation Algorithm Code in MATLAB [111] 46
Sample Code of NN Using MATLAB for Finance Management 48
Required functions [6] 48
Load Historic DAX Prices 48
Plotting Financial Data [6] 48
CAPM [6] 49
Stock Price Prediction Based on Curve Fitting [6] 50
References 51
Part II: Financial Forecasting and Trading 59
2: Trading the FTSE100 Index: ‘Adaptive’ Modelling and Optimization Techniques 60
1 Introduction 60
2 Literature Review 62
3 Related Financial Data 64
4 Proposed Method 66
5 Empirical Results 70
Benchmark Models 70
Trading Performance 71
6 Conclusions and Future Work 77
References 78
3: Modelling, Forecasting and Trading the Crack: A Sliding Window Approach to Training Neural Networks 81
1 Introduction 81
2 Literature Review 86
Modelling the Crack 86
Training of Neural Networks 87
3 Descriptive Statistics 88
4 Methodology 95
The MLP Model 95
The PSO Radial Basis Function Model 97
5 Empirical Results 100
Statistical Accuracy 100
Trading Performance 101
6 Concluding Remarks and Research Limitations 104
7 Appendix 107
Performance Measures 107
Supplementary Information 107
ARMA Equations and Estimations 109
GARCH Equations and Estimations 111
PSO Parameters 116
Best Weights over the Training Windows 116
References 116
4: GEPTrader: A New Standalone Tool for Constructing Trading Strategies with Gene Expression Programming 119
1 Introduction 119
2 Literature Review 120
Genetic Programming and Its Applications to Financial Forecasting 120
Gene Expression Programming and Previous Applications 121
3 Dataset 122
4 GEPTrader 123
Proposed Algorithm 123
GEPTrader Graphical User Interface 125
5 Empirical Results 126
Benchmark Models 126
Statistical Performance 127
Trading Performance 127
6 Conclusions 130
References 131
Part III: Economics 134
5: Business Intelligence for Decision Making in Economics 135
1 Introduction 135
2 Literature Review 137
General Equations for Macroeconomic Output 140
3 Methodology for Creating the  Business-­Automated Data Economy Model 143
4 Empirical Results of the Model 150
5 Conclusions 163
References 166
Part IV: Credit Risk and Analysis 169
6: An Automated Literature Analysis on Data Mining Applications to Credit Risk Assessment 170
1 Introduction 170
2 Materials and Methods 172
Search Criteria 172
Text Mining 173
Topics of Articles 175
Proposed Approach 176
3 Results and Analysis 178
Articles 178
Text Mining 179
Topics of Articles 181
4 Conclusions 183
References 184
7: Intelligent Credit Risk Decision Support: Architecture and Implementations 187
1 Introduction 187
2 Literature Review 188
Machine Learning Techniques 188
Techniques for Classification 190
Credit Risk Problems, Solved by Artificial Intelligence 192
3 Decision Support and Expert Systems for Credit Risk Domain 195
Decision Support Systems: Definitions, Goals, Premises 196
Main Types of Decision Support Systems 197
Recent Developments in Decision Support Systems for Banking Problems 200
Requirements for Credit Risk DSS 201
Financial Standards Based Decision Support 204
Developed Architecture for XBRL-Integrated DSS 207
4 Conclusions 211
References 213
8: Artificial Intelligence for Islamic Sukuk Rating Predictions 219
1 Introduction 219
2 Literature Review 221
What Is Sukuk 221
Sukuk Rating Methodology Based on Recourse of the Underlying Asset 223
Previous Studies on Rating Prediction 224
Variable Selection 226
3 Data and Research Method 227
Data and Sample Selection 227
Dependent and Independent Variables 227
Research Method 228
Multinomial Logit Regression 228
Decision Tree 229
Artificial Intelligence Neural Network 229
4 Result and Analysis 231
Data Screening 231
Multinomial Logistic Result 232
Decision Tree and Artificial Intelligence Neural Network Result 236
Phase one: General training 237
Phase Two: Validation test 239
Result comparison 241
5 Conclusion 243
6 Appendices 244
Appendix 1 244
Appendix 2 245
References 246
Part V: Portfolio Management, Analysis and Optimisation 250
9: Portfolio Selection as a Multi-period Choice Problem Under Uncertainty: An Interaction-Based Approach 251
1 Introduction 251
2 The Model 254
Agents 254
Securities and Portfolios 260
Data 261
3 Simulation Results 262
Baseline Framework 263
Portfolio Selection in a Bear Market 267
Portfolio Selection in a Bull Market 269
4 Consistency in Selection 270
Coefficient of Variation 271
Monte Carlo 273
5 Discussion 276
6 Conclusion 285
Appendix: Fragmented pseudo-code 286
References 287
10: Handling Model Risk in Portfolio Selection Using Multi-Objective Genetic Algorithm 291
1 Introduction 291
2 Portfolio Optimization and Modern Portfolio Theory 292
3 The Concepts of Model Risk 294
4 Multi-Objective Genetic Algorithms for Portfolio Optimization 296
5 A Portfolio’s Sharpe Ratio Error 300
6 Stock Forecasting Models 303
7 The Experiment 304
8 Empirical Results and Analyses 306
9 Conclusions 314
References 316
11: Linear Regression Versus Fuzzy Linear Regression: Does it Make a Difference in the Evaluation of the Performance of Mutual Fund Managers? 317
1 Introduction 317
2 Methodology 319
Treynor-Mazuy model 320
Henriksson-Merton model 321
Fuzzy Linear Regression 322
3 Data Set Description 325
4 Empirical Application 326
Results and Discussion 326
The Performance of Mutual Funds Managers 332
Fuzzy Similarity Ratios 335
5 Conclusions and Future Perspectives 336
References 338
Index 342