Online Portfolio Selection Based on Follow-the-Loser Algorithms

Document Type : Research Paper


1 Msc., Department of Financial Engineering, Faculty of Industrial Engineering, K.N. Toosi University of Technology, Tehran, Iran.

2 Associate Prof., Financial Engineering, Faculty of Industrial Engineering, K.N. Toosi University of Technology, Tehran, Iran.

3 Msc., Department of Financial Engineering, Faculty of Financial Sciences, Kharazmi University, Tehran, Iran.


Objective: Nowadays, the volume and speed of transactions in financial markets have grown dramatically and it is hard to track market changes by using traditional methods. Besides the efficiency of traditional methods, the low speed of these approaches is one of the most important deficiencies of them because they cannot adapt to high speed of transactions. To overcome this shortcoming, algorithmic trading techniques have been proposed which online portfolio selection is one of the most important of these techniques. So, the purpose of this research is to propose a new algorithm for online portfolio selection which leads to high risk-adjusted return and speeds up the process of portfolio selection.
Methods: In this research, two algorithms have been proposed using multi-period mean reversion which is the basis of follow-the-loser algorithms. In these algorithms, a set of various experts predict the price relative vector of next period. Then, one of existing algorithms in prediction theory with expert advice is used to assign weights to experts. Then, a learning technique is used for portfolio optimization which leads to portfolio of next period.
Results: The results show the superiority of the proposed algorithms to other algorithms existing in literature based on return and risk-adjusted return criteria.
Conclusion: The concept of mean reversion can be better expressed by using multi-period mean reversion. In addition, using different experts’ advices make predictions more accurate and therefore better portfolios are suggested. Also, the use of weighting system indirectly brings robustness in the algorithms because it reduces the weights assigned to experts with poor predictions and transforms it to other experts with proper predictions.


Abdi, M. (2017). Online Portfolio Selection Based on Pattern Matching Algorithms. M. S. Thesis. Faculty of Industrial Engineering, K.N.Toosi University of Tehran. (in Persian)
Agarwal, A., Hazan, E., Kale, S. & Schapire, R. E. (2006). Algorithms for portfolio management based on the newton method. Proceedings of International Conference on Machine Learning, Pennsylvania, USA.
Akcoglu, K., Drineas, P. & Kao, M. Y. (2005). Fast universalization of investment strategies. SIAM Journal on Computing, 34(1), 1–22.
Balvers, R., Wu, Y. & Gilland, E. (2002). Mean Reversion across National Stock Markets and Parametric Contrarian Investment Strategies. The Journal of Finance, 2(3), 745–772.
Blum, A. & Kalai, A. (1999). Universal portfolios with and without transaction costs. Machine Learning, 35(3), 193–205.
Borodin, A., El-Yaniv, R. & Gogan, V. (2004). Can we learn to beat the best stock. Journal of Artifical Intelligence Research, 21(1), 579–594.
Cover, T. M. & Ordentlich, E. (1996). Universal portfolios with side information. IEEE Transactions on Information Theory, 42(2), 348–363.
Cover, T. M. & Ordentlich, E. (1998). Universal portfolios with short sales and margin. In Proceedings of Annual IEEE International Symposium on Information Theory, Cambridge, MA.
Cover, T. M. (1991). Universal portfolios. Mathematical Finance, 1(1), 1-29.
Crammer, K., Dekel, O., Keshet, J., Shalev-Shwartz, S. & Singer, Y. (2006). Online passive aggressive algorithms, Journal of Machine Learning Research, 7(3), 551–585.
Crammer, K., Dredze, M. & Pereira, F. (2008). Exact convex confidence weighted learning. Advances in Neural Information Processing Systems 21, 345–352.
Cross, J. E. & Barron, A. R. (2003). Efficient universal portfolios for past-dependent target classes. Mathematical Finance, 13(2), 245–276.
Cunado, J., Gil-Alana, L. A. & de Gracia, F. P. (2010). Mean reversion in stock market prices: New evidence based on bull and bear markets. Research in International Business and Finance 24, 113–122.
Das, P. & Banerjee, A. (2011). Meta optimization and its application to portfolio selection. In Proceedings of the 17th ACM SIGKDD international conference on Knowledge discovery and data mining, California, USA, 1163-1171.
DeBondt, W. F. and Thaler, R. (1985). Does the stock market overreact? Journal of Finance, 40(3), 793–805.
Duchi, J., Shalev-Shwartz, S., Singer, Y. & Chandra, T. (2008). Efficient projections onto the l1-ball for learning in high dimensions. In Proceedings of International Conference on Machine Learning, Helsinki, Finland, 272–279.
Fama, E. F. & French, K. R. (1988). Permanent and temporary components of stock prices. Journal of Political Economy, 96, 246–273.
Freund, Y. & Schapire, R. E. (1997). A Decision-Theoretic Generalization of On-Line Learning and an Application to Boosting. Journal of Computer and System Sciences, 55(1), 119–139.
Gaivoronski, A. A. & Stella, F. (2000). Stochastic nonstationary optimization for finding universal portfolios. Annals of Operations Research 100(1), 165–188.
Gaivoronski, A. A. & Stella, F. (2003). On-line portfolio selection using stochastic programming. Journal of Economic Dynamics and Control, 27(6), 1013–1043.
Gao, L., Zhang, W. & Tang, Q. (2013). Passive Aggressive Algorithm for Online Portfolio Selection with Piecewise Loss Function. Springer-Verlag Berlin Heidelberg, 360–371.
Györfi, L. & Schäfer, D. (2003). Nonparametric prediction. In Advances in Learning Theory: Methods, Models and Applications, Amsterdam, The Netherlands, 339–354.
Györfi, L., Lugosi, G. & Udina, F. (2006). Nonparametric kernel-based sequential investment strategies. Mathematical Finance, 16(2), 337–357.
Györfi, L., Udina, F. & Walk, H. (2008). Nonparametric nearest neighbor based empirical portfolio selection strategies. Statistics and Decisions, 26(2), 145–157.
Györfi, L., Urban, A. & Vajda, I. (2007). Kernel-based semi-log-optimal empirical portfolio selection strategies. International Journal of Theoretical and Applied Finance, 10(3), 505–516.
Hazan, E. & Seshadhri, C. (2009). Efficient learning algorithms for changing environments. In Proceedings of the International Conference on Machine Learning, Montreal, Canada, 393-400.
Helmbold, D., Schapire, R., Singer, Y. & Warmuth, M. (1998). On-line portfolio selection using multiplicative updates. Mathematical Finance, 8(4), 325–347.                   
Huang, D., Zhou, J., Li, B., Hoi, S. C. & Zhou, S. (2013). Robust median reversion strategy for on-line portfolio selection. In Proceedings of the 23th International Joint Conference on Artificial Intelligence, 2006–2012.
Jegadeesh, N. (1991). Seasonality in Stock Price Mean Reversion: Evidence from the U.S. and the U.K. The Journal of Finance, 4(2), 1427–1444.
Karbasi, H., Nouri, Y. & Chenari, H. (2012). Study on mean reversion phenomenon in Tehran Stock Exchange using unit root test. Journal of Investment Knowledge, 1(4), 87-103. 
(in Persian)
Li, B. & Hoi, S. C. H. (2012). On-line portfolio selection with moving average reversion. In Proceedings of the 29th International Conference on Machine Learning, Edinburgh, Scotland, 1–8.
Li, B. & Hoi, S. C. H. (2015). Online Portfolio Selection: Principles and Algorithms. New York, CRC Press.
Li, B., Hoi, S. C. H. & Gopalkrishnan, V. (2011). Corn: Correlation-driven nonparametric learning approach for portfolio selection. ACM Transactions on Intelligent Systems and Technology,2(3), 1–29.
Li, B., Hoi, S. C. H., Zhao, P. & Gopalkrishnan, V. (2013). Confidence weighted mean reversion strategy for online portfolio selection. ACM Transactions on Knowledge Discovery from Data,7(1), 1–38.
Li, B., Zhao, P., Hoi, S. C. H. & Gopalkrishnan, V. (2012). PAMR: Passive aggressive mean reversion strategy for portfolio selection. Machine Learning,87(2), 221–258.
Poterba, J. & Summers, L. (1988). Mean reversion in stock prices: Evidence and implications, Journal of Financial Economics, 22(1), 27–59.
Serletis, A. & Rosenberg, A. A. (2007). Mean reversion in the US stock market. Chaos, Solitons and Fractals, 40(4), 2007–2015.
Tehrani, R., Ansari, H. & Sardan, A. (2008). Investigation on mean reversion phenomenon in Tehran Stock Exchange using variance ratio test. Journal of Accounting and Auditing reviews, 15(54), 17-32. (in Persian)
Vovk, V. G. & Watkins, C. (1998). Universal portfolio selection. In Proceedings of the 11th annual Conference on Learning Theory, Wisconsin, USA, 12–23.
Wang, Y., Wang, D., Wang, Y. & Zhang, Y. (2018). Racorn-K: Risk-Aversion Pattern Matching-Based Portfolio Selection. Quantitative Finance > Risk Management [online]. Available from: [accessed 28 Feb 2018].
Zijin, p. (2016). On-Line Portfolio Selection Strategy Based on Weighted Moving Average Asymmetric Mean Reversion. Management Science and Engineering, 10(1), 43-48.