인샘플 탄젠시 포트폴리오의 구성을 이용한 포트폴리오 연구의 새로운 접근

This paper studies portfolio in a compositional perspective and presents 1) the evaluation method of portfolio composition and 2) the mechanism of portfolio forecasting. The evaluation method of portfolio composition quantifies portfolio’s compositional difference with the in-sample tangency portfolio using the vector distance. In attempts to select the most suitable method of evaluating portfolio composition, four distance methods are compared, which are categorized into the norm and the inner-product distances. Then, regression and distribution analyses demonstrate that Euclidean distance can be used as a new measure to evaluate the superiority of portfolios. The mechanism of portfolio forecasting tracks the past in-sample tangency portfolios and finds a portfolio that is similar to the tangency portfolio at the time of investment. Through numerical experiments, we analyze the forecasted portfolio from the mechanism applying SMA and EWMA, and then confirm empirical significance of the composition-based portfolio research.