Statistical Methods in Finance 2022

June 28 to July 2, 2022


Norm constrained minimum variance portfolios with short selling

By Vrinda
IIT, Roorke

One of the essential features of an efficient market is to reflect the composite viewpoint of all competing interests. To admit only the view focusing on higher prices provides a one-sided perspective of the market while compiling an open expression for both the long and short positions furnishes a two-sided market view that better reflects prevailing conditions of the price structure. In this paper, we attempt to model short selling in the minimum variance portfolio model to take advantage of both the price structures. To exploit the attractiveness of short selling in the model, we propose to utilize the short rebate gain in its objective function. This factor helps in choosing the assets to short based on declining price trends and maximizing the portfolio return. Further, we use respective 1-norm and 2-norm constraints in the proposed model to assimilate the investor’s choice for sparse and diversified portfolios. We present empirical results highlighting the effect of the specific choice of budget constraint and the impact of including the short rebate term in the model. We then conduct a numerical comparative analysis of our proposed models vis-a-vis four comparable models from literature across five global data sets using the rolling window scheme. We observe a justified distribution of the budget in long and short from the proposed models, abstaining from the dominance of one over the other. Further, we find that the proposed models generate better out-of-sample results in comparison to other models on the basis of several important financial performance indices.