Abstract
This project is aim to solve the problem of selecting ARP. Thereby, I will introduce optimization model.
ARP is a financial portfolio that aims to produce a good return regardless of how the underlying market performs. They are sometimes called market neutral portfolios as they are designed to have a low correlation with overall market return.
In this model formation, I adopt the view that in seeking for ARP, we are seeking for a portfolio that achieves a constant return per time period. Obviously, we may not find such portfolio with this property. But in terms of what we desire, my view is that if we can find such a portfolio, then we would have an ideal ARP. With this, the notion that an ARP is somehow `disconnected' from the market is captured by the constancy of return. Therefore, ARP in this work is defined as a portfolio with constant return per time period.
In addition to my view that the desire portfolio should have a constant return, the return should not be specified. Specifying target return might lead to inefficiency in the optimization. Instead, it takes the return that emerges as a result of an optimization.
I intend to present a definition via a three- stage mixed-integer zero-one program for the problem of selecting an ARP. The first stage will be the regression slope; the second will be regression intercept while the third will be the transaction cost. This will be extended to ARP-RT, ARP-ERT and ARP-RERT.
I will present the computational result of the three models (ARP-RT, ARP-ERT and ARP-RERT) using python package for optimization. I intend to use weekly data taken from the universe of assets defined by the S&P global 1200 index and subindices over a period of ten years. Source of data will be from Google finance.
Also, I will set up a hypothesis test for the models before drawing out conclusion.
Key word: ARP-RT – ARP base on the regression of return against time, ARP-ERT – ARP base on the regression of excess return against time, ARP-RERT – ARP base on the regression of return and excess return against time.