Speaker: Dr. Zhihua Zhao
Time: 16:00-17:00, Mar. 5
Location: SIST 1A-200
Host: Prof. Hao Wang
Abstract:
Institutional investors rebalance their portfolios regularly to control the risk pro le of their holdings. We propose a new portfolio rebalancing method with embedded sparsity and robustness, two highly demanded properties by portfolio managers. Sparsity, quantified by the l_0 norm, has the desirable property to trade only a small number of assets in the portfolio and hence potentially reduce trading cost and improve out-of-sample performance. Robustness, achieved by the ellipsoidal uncertainty set for asset returns, has the important property of mitigating estimation errors. Assuming diagonal covariance matrices, we derive a closed-form solution to our problem. This analytical solution allows us to identify two key factors in portfolio rebalancing, namely, the consistent convergence of optimal portfolio positions and the movement of the trading critical points. For non-diagonal covariance matrices, we propose an efficient Alternating Direction Method of Multipliers, where each sub-problem admits a closed-form solution. To develop further intuition and insights, we illustrate these findings via simulated data sets and conduct out-of-sample performance analysis of the proposed method on actual data from the Chinese stock market
Bio:
Dr. Zhihua Zhao, currently serving as a PhD student in the School of Economics and Finance, Xi'an Jiaotong University. His main research interests are quantitative finance, statistics, sparse optimization theory and algorithms involved in big data and microscopic research on typical financial issues. His publications have appeared in famous journals, including Neural Computing & Applications, Science China Mathematics, and Journal of the Operational Research Society etc. His research has been recognized by numerous awards and supported by two National Natural Science Foundation of China as well as a Key Program of the NSF. He has been a visiting scholar in Academy of Mathematics and Systems Science, Chinese Academy of Sciences, and Xidian University ect
SIST-Seminar 18130
沪公网安备 31011502006855号
