报告内容摘要
It is fair to say that stock market prices are believed to evolve according to a random walk and thus cannot be predicted. On the other hand, in some stock markets, we can observe that there are pairs of stocks whose spreads tend to stay around some deterministic level, leading to the idea of “cointegration of pairs of stocks.” If a pair of stocks is cointegrated, their spread is mean reverting, and we can take advantage of this property to construct a portfolio of multiple spreads of cointegrated pairs.

In this work, we formulate a portfolio optimization problem of the spreads of cointegrated pairs of stocks, and develop a model predictive control (MPC) approach based on a conditional mean-variance (MV) optimization with a given prediction horizon. To this end, we first express the spread processes as a vector autoregressive (VAR) model, and show that the conditional MV optimal portfolio may be computed efficiently given information up to the current time. Based on the solution to the conditional MV problem, we apply an MPC strategy that calculates the conditional MV optimal portfolio for a given prediction horizon at each step.  Here we demonstrate empirical simulations using stock price data consisting of Nikkei 225 in Japan, where we apply our MPC approach to construct optimal portfolios, and evaluate the out-of-sample performance with transaction costs for different values of prediction horizons and rebalance intervals. Note that this work is under collaboration with Prof. Primbs at the California State University, Fullerton, USA.

---

报告人简介

Yuji Yamada is a professor in the Faculty of Business Sciences, University of Tsukuba, Japan.  He holds an undergraduate degree in Engineering from University of Chiba, a Masters degree and PhD in Engineering from Tokyo Institute of Technology. Before joining University of Tsukuba, he served as a postdoctoral scholar at Control and Dynamical Systems, Caltech.  At University of Tsukuba, he teaches courses in Mathematical Finance and Financial Engineering.  His current research interests are at the intersection between finance, optimization, and control theory.

 

欢迎广大师生光临！