If one is going to replicate the Yale Portfolio, as suggested in this article, what ETFs meet the highest standards based on "Cluster Weighting Momentum" analysis? To answer this question, we take the eighteen (18) ETFs suggested in the article and run a ranking and cluster analysis to find those ETFs that carry the lowest correlation and highest performance.
ETF Rankings: The following data table ranks the eighteen "Yale Replication" ETFs plus SPY and SHY. SPY is included as a reference to the broad U.S. Equities market and SHY is the cutoff ETF. ETFs that are not outperforming SHY are eliminated from the portfolio until they once more show gains in excess of the cutoff SHY.
The ranking is based on three factors. Fifty percent (50%) of the total weight is assigned to the performance over the most recent three months (91 days). Thirty percent (30%) is allocated to the performance over the most recent six months (182 days) and the final 20% is apportioned to ETF volatility. The table also includes additional technical data such as several Exponential Moving Averages, the "Golden Cross" (X/O) and Absolute Acceleration. In this analysis we will concentrate on the ranking with special attention to ETFs and their performance position with respect to SHY.
Cluster Analysis: The following cluster analysis is designed to find ETFs that have a low correlation with each other. The cutoff was selected to be 0.80 or 80% so as to create 10 clusters. For example, IGE and XOP carry a correlation of 0.86 and are considered highly correlated with each other. Neither is performing better than SHY so they are not recommended for inclusion in the current portfolio. BX carries a correlation of 0.55, occupies a cluster by itself, and is performing above SHY so it is included in the portfolio.
This cluster analysis breaks the 18 ETFs into ten clusters. ETFs that are the top performer in each cluster is a candidate for portfolio inclusion, provided the ETF is also outperforming SHY.
Buy-Hold-Sell Recommendations: Four ETFs made the final cut and they are: PSP, BX, JNK, and EVR. If one is setting up a $100,000 portfolio the number of shares is recommended in the following table. This is based on data from 1/24/2014.
Cluster Rankings and Weightings: The following table extracts data from the above graph, simplifying the information. No ETFs from the first six clusters made the cut as all are performing below the SHY cutoff ETF.
This type of analysis ends up concentrating investments in a few ETFs, not a suitable option for all investors. Due to the low correlations, these few ETFs end up building a well-diversified portfolio. It is not the number of securities that brings diversification, but rather the correlation among those securities.
Disclosure: I am long JNK, VEU, VNQ. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.