Evolving robust GP solutions for hedge fund stock selection in emerging markets

ABSTRACT

Stock selection for hedge fund portfolios is a challenging problem for Genetic Programming (GP) because the markets (the environment in which the GP solution must survive) are dynamic, unpredictable and unforgiving. How can GP be improved so that solutions are produced that are robust to non-trivial changes in the environment? We explore an approach that uses subsets of extreme environments during training.

REFERENCES

• 1: V. S. Aragon and S. C. Esquivel. An evolutionary algorithm to track changes of optimum value locations in dynamic environments. Journal of Computer Science and Technology 4(3):127--134, 2004.
• 2: T. F. Bersano-Begey and J. M. Daida. A discussion on generality and robustness and a framework for fitness set construction in genetic programming to promote robustness. In J. R. Koza, editor, Late Breaking Papers at the 1997 Genetic Programming Conference pp. 11--18, Stanford Bookstore, 1997.
• 3: C. P. Bowers. Formation of modules in a computational model of embryogeny. In The 2005 IEEE Congress on Evolutionary Computation volume 1, pages 537--542, 2005.
• 4: Jürgen Branke, Creating Robust Solutions by Means of Evolutionary Algorithms, Proceedings of the 5th International Conference on Parallel Problem Solving from Nature, p.119-128, September 27-30, 1998
• 5: J. Michael Fitzpatrick , John J. Grefenstette, Genetic Algorithms in Noisy Environments, Machine Learning, v.3 n.2-3, p.101-120, October 1988 [doi>10.1007/BF00113893]
• 6: C. Gagné, M. Schoenauer, M. Parizeau, and M. Tomassini. Genetic programming, validation sets, and parsimony pressure. In Proceedings of the 9th European Conference on Genetic Programming LNCS 3905, pp. 109--120, Springer, 2006.
• 7: A. Gierer, S. Berking, H. Bode, C. N. David, K. Flick, G. Hansmann, H. Schaller, and E. Trenkner. Regeneration of hydra from reaggregated cells. Nature New Biology 239:98--101, 1972.
• 8: Ulrich Hammel , Thomas Bäck, Evolution Strategies on Noisy Functions: How to Improve Convergence Properties, Proceedings of the International Conference on Evolutionary Computation. The Third Conference on Parallel Problem Solving from Nature: Parallel Problem Solving from Nature, p.159-168, October 09-14, 1994
• 9: Thomas Haynes , Roger L. Wainwright , Sandip Sen , Dale A. Schoenefeld, Strongly Typed Genetic Programming in Evolving Cooperation Strategies, Proceedings of the 6th International Conference on Genetic Algorithms, p.271-278, July 15-19, 1995
• 10: J. Herrmann. A genetic algorithm for minmax optimisation problems. volume 2, pp. 1099--1103, 1999.
• 11: M. A. Huynen, P. F. Stadler and W. Fontana, Smoothness within ruggedness: The role of neutrality in adaptation. Proceedings of the National Academy of Sciences of the United States of America (PNAS) 93(1):397--401, 1996.
• 12: T. Ito, H. Iba, and M. Kimura. Robustness of robot programs generated by genetic programming. In Genetic Programming 1996: Proceedings of the First Annual Conference MIT Press, 1996. 321--326.
• 13: E. Jordaan, A. Kordon, L. Chiang, and G. Smits. Robust inferential sensors based on ensemble of predictors generated by genetic programming. In Parallel Problem Solving from Nature - PPSN VIII LNCS 3242, pp. 522--531, Springer-Verlag, 2004.
• 14: H. Kitamo. Foundations of systems biology MIT Press, ISBN 0-262-11266-3, 2001.
• 15: Ibrahim Kuscu, Promoting Generalisation of Learned Behaviours in Genetic Programming, Proceedings of the 5th International Conference on Parallel Problem Solving from Nature, p.491-500, September 27-30, 1998
• 16: I. Kuscu. Generalisation and domain specific functions in genetic programming. In Proceedings of the 2000 Congress on Evolutionary Computation CEC00 Vol. 2, pp. 1393--1400, IEEE Press, 2000.
• 17: I. Kushchu. Genetic programming and evolutionary generalization. IEEE Transactions on Evolutionary Computation 6(5):431--442, October 2002.
• 18: R. Lowenstein. When Genius Failed Fourth Estate, 2002.
• 19: B. L. Miller and D. E. Goldberg. Genetic algorithms, selection scheme, and the varying effect of noise. Evolutionary Computation 4(2):113--131, 1996.
• 20: J. F. Miller. Evolving a self-repairing, self-regulating, french flag organism. pages 129--139, 2004.
• 21: F. W. Moore and O. N. Gacia. A new methodology for reducing brittleness in genetic programming. In Proceedings of the National Aerospace and Electronics 1997 Conferences, NAECON-97 1997.
• 22: 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), 1998.
• 23: L. Panait and S. Luke. Methods for evolving robust programs. In Genetic and Evolutionary Computation - GECCO 2003 volume 2724 of LNCS pages 1740--1751. Springer, 2003.
• 24: Craig W. Reynolds, An evolved, vision-based behavioral model of coordinated group motion, Proceedings of the second international conference on From animals to animats 2 : simulation of adaptive behavior: simulation of adaptive behavior, p.384-392, August 1993, Honolulu, Hawai, United States
• 25: J. Rosca. Generality versus size in genetic programming. In Genetic Programming 1996: Proceedings of the First Annual Conference pp. 381--387. MIT Press, 1996.
• 26: W. F. Sharpe. The sharpe ratio. J. Portfolio Management 21:49--58, 1994.
• 27: T. Soule. Operator choice and the evolution of robust solutions. In R. L. Riolo and B. Worzel, editors, Genetic Programming Theory and Practise chapter 16, pages 257--270. Kluwer, 2003.
• 28: T. Soule, R. B. Heckendorn, and J. Shen. Solution stability in evolutionary computation. In ISCIS XVII International Symposium On Computer and Information Sciences pp. 237--241, CRC Press, 2002.
• 29: S. Tsutsui and A. Ghosh. Genetic algorithms with a robust solution searching scheme. IEEE Transactions on Evolutionary Computation 1(3):201--208, 1997.
• 30: A. Wagner. Robustness and Evolvability in Living Systems Princeton University Press, 2005.

INDEX TERMS

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

General Terms:
Algorithms, Experimentation

Keywords:
adaptation, diversity, dynamic environment, finance, genetic programming, phenotype