Learning to optimize profits beats predicting returns -: comparing techniques for financial portfolio optimisation

ABSTRACT

Stock selection for hedge fund portfolios is a challenging problem that has previously been tackled by many machine-learning, genetic and evolutionary systems, including both Genetic Programming (GP) and Support Vector Machines (SVM). But which is the better? We provide a head-to-head evaluation of GP and SVM applied to this real-world problem, including both a standard comparison of returns on investment and a comparison of both techniques when extended with a "voting" mechanism designed to improve both returns and robustness to volatile markets. Robustness is an important additional dimension to this comparison, since the markets (the environment in which the GP or SVM solution must survive) are dynamic and unpredictable.

Our investigation highlights a key difference in the two techniques, showing the superiority of the GP approach for this problem.

REFERENCES

• 1: E. Alfaro-Cid, K. Sharman, and A. Esparcia-Alcazar. A genetic programming approach for bankruptcy prediction using a highly unbalanced database. In EvoWorkshops, volume 4448 of LNCS, pages 169--178. Springer, 2007.
• 2: K.-Y. Chen and C.-H. Ho. An improved support vector regression modeling for Taiwan Stock Exchange market weighted index forecasting. In International Conference on Neural Networks and Brain, ICNN&B '05., volume 3, pages 1633--1638. IEEE, 2005.
• 3: Thomas G. Dietterich, Ensemble Methods in Machine Learning, Proceedings of the First International Workshop on Multiple Classifier Systems, p.1-15, June 21-23, 2000
• 4: A. Fan and M. Palaniswami. Stock selection using support vector machines. In Proc. International Joint Conference on Neural Networks, IJCNN '01., volume 3, pages 1793--1798. IEEE, 2001.
• 5: A. A. Gaivoronski, S. Krylov, and N. van der Wijst. Optimal portfolio selection and dynamic benchmark tracking. European Journal of Operational Research, 163(1):115--131, Elsevier, 2005.
• 6: J. Herrmann. A genetic algorithm for minmax optimisation problems. In Proc. of the Congress on Evolutionary Computation, volume 2, pages 1099--1103. IEEE, 1999.
• 7: J. Kittler and F. Roli. Multiple classifier systems. In Proc. of MCS 2001. Springer Verlag, 2001.
• 8: H. M. Markowitz. Portfolio selection: efficient diversification of investment. MIT Business Review, 3(1):120--125, 1959.
• 9: J. F. Miller. Evolving a self-repairing, self-regulating, french flag organism. volume 3102 of LNCS, pages 129--139. Springer, 2004.
• 10: F. W. Moore and O. N. Gacia. A new methodology for reducing brittleness in genetic programming. In Proc. of the National Aerospace and Electronics 1997 Conferences, NAECON-97. IEEE, 1997.
• 11: V. Nissen and J. Propach. On the robustness of population-based versus point-based optimisation in the presence of noise. IEEE Transactions on Evolutionary Computation, 2(3):107--119, 1998.
• 12: K. J. Oh, T. Y. Kim, and S. Min. Using genetic algorithm to support portfolio optimization for index fund management. Expert Systems with Applications, 28(2):371--379, Elsevier, 2005.
• 13: Sancho Salcedo-Sanz , José-Luis Fernández-Villacañas , María Jesús Segovia-Vargas , Carlos Bousoño-Calzón, Genetic programming for the prediction of insolvency in non-life insurance companies, Computers and Operations Research, v.32 n.4, p.749-765, April 2005 [doi>10.1016/j.cor.2003.08.015]
• 14: C. Schoreels and J. M. Garibaldi. The effect of varying parameter on performance for adaptive agents in techical equity market trading. In Proc. of the 3rd IEEE Internatioal Conference on Computational Cybernetics (ICCC 2005), 2005.
• 15: W. F. Sharpe. The Sharpe Ratio. J. Portfolio Management, 21:49--58, 1994.
• 16: T. Soule. Voting teams: A cooperative approach to non-typical problems using genetic programming. In Proc. of the Genetic and Evolutionary Computation Conference, volume 1, pages 916--922. Morgan Kaufmann, 1999.
• 17: T. Soule. Operator choice and the evolution of robust solutions. In R. L. Riolo and B. Worzel, editors, Genetic Programming Theory and Practice, chapter 16, pages 257--270. Kluwer, 2003.
• 18: Vladimir N. Vapnik, The nature of statistical learning theory, Springer-Verlag New York, Inc., New York, NY, 1995
• 19: A. S. Vieira, B. Ribeiro, S. Mukkamala, J. C. Neves, and A. H. Sung. On the performance of learning machines for bankruptcy detection. In Proc. 2nd IEEE International Conference on Computational Cybernetics, ICCC'04., pages 323--327. IEEE, 2004.
• 20: L. Wagman. Stock portfolio evaluation: An application of genetic-programming-based technical analysis. In J. R. Koza, editor, Genetic Algorithms and Genetic Programming at Stanford 2003, pages 213--220. Stanford Bookstore, Stanford, California, 94305-3079 USA, 2003.
• 21: Hsing-Wen Wang, Portfolio selection with fuzzy MCDM using genetic algorithm: application of financial engineering, Proceedings of the 17th IASTED international conference on Modelling and simulation, p.597-602, May 24-26, 2006, Montreal, Canada
• 22: Wei Yan , Christopher D. Clack, Diverse committees vote for dependable profits, Proceedings of the 9th annual conference on Genetic and evolutionary computation, July 07-11, 2007, London, England [doi>10.1145/1276958.1277383]
• 23: Wei Yan , Christopher D. Clack, Evolving robust GP solutions for hedge fund stock selection in emerging markets, Proceedings of the 9th annual conference on Genetic and evolutionary computation, July 07-11, 2007, London, England [doi>10.1145/1276958.1277384]
• 24: Byoung-Tak Zhang , Je-Gun Joung, Building Optimal Committees of Genetic Programs, Proceedings of the 6th International Conference on Parallel Problem Solving from Nature, p.231-240, September 18-20, 2000
• 25: Defu Zhang , Hongyi Huang , Qingshan Chen , Yi Jiang, A Comparison Study of Credit Scoring Models, Proceedings of the Third International Conference on Natural Computation (ICNC 2007), p.15-18, August 24-27, 2007 [doi>10.1109/ICNC.2007.15]
• 26: C. Zhou, L. Yu, T. Huang, S. Wang, and K. K. Lai. Selecting valuable stock using genetic algorithm. In Proc. SEAL'06, pages 688--694. Springer, 2006.

INDEX TERMS

Primary Classification:
  I. Computing Methodologies
  I.2 ARTIFICIAL INTELLIGENCE
      I.2.m Miscellaneous

General Terms:
Algorithms, Experimentation

Keywords:
GP, SVM, committee, diversity, dynamic environment, finance, robustness, voting