Improving Portfolio Performance Using Attribute Construction and Combinatorial Fusion
In portfolio management, stocks can be evaluated based on its financial attributes. Recent research has shown that attribute construction and combination using combinatorial fusion analysis can achieve better results than individual attributes. In this thesis, two weighted combination methods are proposed and implemented: weighted combination by recency (AR) and weighted combination by discounting (AD). Approach AR gives a linear increment on the weight of the data while approach AD gives an exponential increment on the weight of the data. This thesis demonstrates that the two weighted combination methods can outperform the simple average combination method. They both perform better than the Russell 2000 index in the test case.
Wang, Xiaoran, "Improving Portfolio Performance Using Attribute Construction and Combinatorial Fusion" (2019). ETD Collection for Fordham University. AAI22615670.