Abstract
We propose an iterative gradient descent algorithm for solving scenario-based Mean-CVaR portfolio selection problem. The algorithm is fast and does not require any LP solver. It also has efficiency advantage over the LP approach for large scenario size.
Similar content being viewed by others
References
Agarwal, V., & Naik, N. (2004). Risks and portfolio decisions involving hedge funds. The Review of Financial Studies, 17(1), 63–98.
Alexander, S., Coleman, T., & Li, Y. (2006). Minimizing CVaR and VaR for a portfolio of derivatives. Journal of Banking & Finance, 30(2), 583–605.
Andersen, E. D., & Andersen, K. D. (2006). The MOSEK optimization toolbox for MATLAB manual Version 4.0. http://www.mosek.com/products/4_0/tools/help/index.html.
Angelelli, E., Mansini, R., & Speranza, M. (2008). A comparison of MAD and CVaR models with real features. Journal of Banking & Finance, 32(7), 1188–1197.
Artzner, P., Delbean, F., Eber, J., & Heath, D. (1999). Coherent measure of risks. Mathematical Finance, 9(3), 203–228.
Basak, S., & Shapiro, A. (2001). Value-at-risk based risk management: optimal policies and asset prices. The Review of Financial Studies, 14(2), 371–405.
Black, F., & Litterman, R. (1990). Asset allocation: combining investor views with market equilibrium. Goldman Sachs Fixed Income Research.
Black, F., & Litterman, R. (1991). Asset allocation: combining investor views with market expectations. The Journal of Fixed Income, 1(1), 7–18.
Chiodi, L., Mansini, R., & Speranza, M. (2003). Semi-absolute deviation rule for mutual funds portfolio selection. Annals of Operations Research, 124(1), 245–265.
Gaivoronski, A., & Pflug, G. (2005). Value-at-risk in portfolio optimization: properties and computational approach. The Journal of Risk, 7(2), 1–31.
ILOG (2008). ILOG CPLEX 11.1. http://www.ilog.com/products/cplex/.
Konno, H., & Yamazaki, H. (1991). Mean-absolute deviation portfolio optimization model and its applications to Tokyo stock market. Management Science, 37(5), 519–531.
Koskosidis, Y., & Duarte, A. Jr. (1997). A scenario-based approach to active asset allocation. The Journal of Portfolio Management, 23, 74–85.
Larsen, N., Mausser, H., & Uryasev, S. (2002). Algorithms for optimization of value-at-risk. In Applied optimization series. Financial engineering, e-commerce and supply chain (pp. 19–46).
Lüthi, H., & Doege, J. (2005). Convex risk measures for portfolio optimization and concepts of flexibility. Mathematical Programming, 104(2), 541–559.
Markowitz, H. (1952). Portfolio selection. The Journal of Finance, 7(1), 77–91.
Mas-Colell, A., Whinston, M., & Green, J. (1995). Microeconomic theory. London: Oxford University Press.
Meucci, A. (2006). Beyond Black-Litterman: views on non-normal markets. Risk Magazine, 19, 87–92.
Nesterov, Y. (2005). Smooth minimization of non-smooth functions. Mathematical Programming, 103(1), 127–152.
Rockafellar, R., & Uryasev, S. (2000). Optimization of conditional value-at-risk. The Journal of Risk, 2(3), 21–41.
Rockafellar, R., & Uryasev, S. (2002). Conditional value-at-risk for general loss distributions. Journal of Banking & Finance, 26(7), 1443–1471.
Rockafellar, R., Uryasev, S., & Zabarankin, M. (2006a). Deviation measures in risk analysis and optimization. Finance and Stochastics, 10(1), 51–74.
Rockafellar, R., Uryasev, S., & Zabarankin, M. (2006b). Optimality conditions in portfolio analysis with general deviation measures. Mathematical Programming, 108(2–3), 515–540.
Sharpe, W. (1971). Mean-absolute-deviation characteristic lines for securities and portfolios. Management Science, 18(2), B1–B13.
Speranza, M. (1993). Linear programming models for portfolio optimization. The Journal of Finance, 14(1), 107–123.
Author information
Authors and Affiliations
Corresponding author
Appendices
Appendix A: Computation of constants
For completeness, we compute the constants here and they can be used for computing the number of iterations required given in (34). Please refer to Nesterov (2005) for the details. First,
To solve min q∈Q d 2(q), we find that
Setting \(\frac{\partial L}{\partial q_{i}} = 0\) gives
Therefore q i =p i and
Note that max q∈Q d 2(q) occurs at extreme points where there are K of q i equal to p i β −1 and N−K of them equal to 0, this gives max q∈Q d 2(q)=∑ i p i β −1log(p i β −1). So we have
As in (Nesterov 2005), σ 2 is defined as the constant such that \(d_{2}(\mathbf{q}) \geq\frac{1}{2} \sigma_{2} \|\mathbf{q} - \mathbf{p}\| _{1}^{2}\). By Cauchy-Schwartz inequality,
Consider
we have \(q_{i} = \frac{(1 - \gamma)\beta^{-1}}{2} p_{i} \propto p_{i}\), and thus q i =p i . As a result,
Finally, by (11)
By setting \(\mu= \frac{\varepsilon}{2 D_{2}}\), we can guarantee that our algorithm stops by at most K iterations as defined in (34).
Appendix B: Algorithm
Rights and permissions
About this article
Cite this article
Iyengar, G., Ma, A.K.C. Fast gradient descent method for Mean-CVaR optimization. Ann Oper Res 205, 203–212 (2013). https://doi.org/10.1007/s10479-012-1245-8
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10479-012-1245-8