Dynamic Adaptive Algorithm Selection: Profit Maximization for Online Trading
Online trading algorithms can facilitate the investment decisions in financial markets. This paper presents a dynamic adaptive algorithm selection framework for online trading with the goal of maximizing overall revenue (lower competitive ratio). We integrate the algorithm selection () and probabilistic graphs to transform the algorithm selection model of offline algorithms for a dynamically adaptive model of online trading. Unlike the traditional static approach of algorithm selection, where a single algorithm is executed over the whole investment horizon, we dynamically select and update the trading algorithm by analyzing the time series features. The time series is partitioned in different windows and each window’s features are extracted sequentially using real time hybrid pattern matching approach. The extracted features of current window w i are analyzed and the decision making module determines the best suitable algorithm for next window w i + 1 on the basis of underlying probabilistic graphical model. The process is repeated until all windows in a time series are processed. Our dynamic adaptive algorithm selection model outperforms the static model on real world datasets of DAX30 and S&P500.
KeywordsBusiness applications of artificial intelligence Business Intelligence Online Trading Adaptive Algorithms
Unable to display preview. Download preview PDF.
- 3.Zhang, Z., Jiang, J., Liu, X., Lau, R., Wang, H., Zhang, R.: A real time hybrid pattern matching scheme for stock time series. In: Proceedings of the Twenty-First Australasian Conference on Database Technologies, ADC 2010, vol. 104, pp. 161–170. Australian Computer Society, Inc., Darlinghurst (2010)Google Scholar
- 4.Potkonjak, M., Rabaey, J.: Algorithm selection: a quantitative computation-intensive optimization approach. In: ICCAD 1994: Proceedings of the 1994 IEEE/ACM International Conference on Computer-Aided Design, pp. 90–95. IEEE Computer Society Press, Los Alamitos (1994)Google Scholar
- 5.Lagoudakis, M.G., Littman, M.L.: Algorithm selection using reinforcement learning. In: Proceedings of the Seventeenth International Conference on Machine Learning, ICML 2000, pp. 511–518. Morgan Kaufmann Publishers Inc., San Francisco (2000)Google Scholar
- 6.Hazan, E., Seshadhri, C.: Adaptive algorithms for online decision problems. Electronic Colloquium on Computational Complexity (ECCC) 14(088) (2007)Google Scholar
- 7.Gagliolo, M., Zhumatiy, V., Schmidhuber, J.: Adaptive online time allocation to search algorithms, pp. 134–143 (2004)Google Scholar
- 8.Xu, L., Hutter, F., Hoos, H.H., Leyton-Brown, K.: Satzilla: portfolio-based algorithm selection for sat. J. Artif. Int. Res. 32, 565–606 (2008)Google Scholar
- 10.Jordan, M.I.: Graphical models. Statistical Science (Special Issue on Bayesian Statistics) 19, 140–155 (2004)Google Scholar