On-Line Portfolio Selection via Machine Learning

About OLPS

OLPS is a research project for studying on-line portofolio selection using state-of-the-art machine learning algorithms. In this task, a portfolio manager is a decision maker, whose goal is to produce a portfolio strategy, aiming to maximize the cumulative wealth. He/she computes the portfolios sequentially. On each period, the manager has access to the sequence of previous price relative vectors. Then, he/she computes a new portfolio for next unknown price relative vector, where the decision criterion varies among different managers. The portfolio is scored based on portfolio period return. This procedure is repeated until the end, and the portfolio strategy is finally scored according to portfolio cumulative wealth.