An Optimal Investment Strategy for University Endowment Fund

Abstract

Automatic educational endowment funding systems suffer from two problems, i.e., lack of historical data and labels. In this paper, we propose a simple but effective investment portfolio model, which helps us to determine which campuses should be invested in, how much money should be distributed, and how long funds should last. Our algorithm first uses the PCA framework to assign each campus a label, and then uses SVMs to learn ranking list of campuses. We further develop a greedy strategy to optimize portfolio. Experiments on the ROC metric indicate the effectiveness of our model.