I’ve been grappling over why the performance of different portfolios varies so widely. In this post I’m identifying several variables that I think contribute to performance differences. While this is not exhaustive list, it focuses on several variables we as money managers have some control.
If one digs back to some of the academic research, you will find papers pointing out the importance of asset allocation. Some of the most important work in this area comes from Ibbotson Associates.
Let’s begin by taking these one at a time.
- Investment Quiver: What securities are included in a portfolio cannot be ignored. For example, suppose one invested only in Amazon, Facebook, Google, and Microsoft when each was launched as a public company. Such a portfolio would have done very well, although we would not consider it all that well diversified. It makes common sense to think the selection of securities for a portfolio makes a difference. In all ITA portfolios we include U.S. Equities, International Equities, and Bonds. These are the “Big Three” asset classes. After including these, we branch out to include other factors or asset classes.
- Asset Allocation: For large endowment funds, asset allocation is a major driver when it comes to populating a portfolio. How far should we deviate from this model? While most ITA portfolios include all the important asset classes, we move from asset class to asset class depending on the performance of each. This movement is not always to the advantage of portfolio performance.
- Investing Model: There are four major investment models available within the Kipling spreadsheet. A fifth model, not part of the Kipling, is to simply invest in one single mutual fund such as VFINX or an ETF such as VT or VTI. The four models available within the Kipling are: Dual Momentum, Heikin-Ashi (HA), Buy-Hold-Sell (BHS), and Linear Progression Projection Convolution (LRPC). Six ITA portfolio are currently following the Dual Momentum Model. An experiment is in progress to compare the performance between the HA, BHS, and LRPC models. Those are the three Carson portfolios. Thus far, the HA model is outpacing the other two in this experiment. A sub-set if these major models involves the look-back periods. A second experiment is in operation where three different look-back periods are used with three Dual Momentum portfolios. They are: Franklin, McClintock, and Pauling.
- Level of Risk: I consider this to be a major variable as money managers may well be more interested in protecting capital than they are in portfolio performance. When more data is available, I want to examine the Jensen’s Alpha percentage for the different ITA portfolios. This risk calculation takes into consideration portfolio risk. The end of December is a good time to make comparisons as we will have more data.
- Review Day: One example of this important variable shows up within the Rutherford portfolio where the investment quiver has varied little since it was launched by in October of 2014. The weekly updates of the Rutherford illustrate the impact of this little discussed variable. Another example is the benefit derived by the Galileo portfolio that was reviewed shortly before the Covid-19 crash. Fortunately, the moved called for shifting from equities to U.S. Treasuries – a fortuitous move.
As you manage your portfolio, think about how these variables impact your portfolio and which you want to emphasize. Are you willing to increase the risk to improve the return? Which investing model is your preferred style? Not mentioned above are Robo Advisor or computer managed portfolios. I track one of these (Schrodinger) on this blog. Thus far I am not enamored with this approach to portfolio management.
If you think of other important variables, post your ideas in the Comments section found below.