Analysis of Cambria ETFs (Part II)
September 17, 2016
Equal-Weighting S&P 500
The previous post in this two-part series covered the three actively-managed products out of the five Cambria ETFs with a sufficiently long history. This post will focus on the remaining two index ETFs.
Let’s start with the analysis of the Cambria Foreign Shareholder Yield ETF (FYLD). According to the sponsor, the fund follows a proprietary index that
…consists of stocks with high cash distribution characteristics. The initial screening universe for this Index includes stocks in foreign developed countries with marketing capitalizations over $200 million. The Index is comprised of the 100 companies with the best combined rank of dividend payments and net stock buybacks, which are the key components of shareholder yield. The Index also screens for value and quality factors, including low financial leverage.
As in the case of actively-managed Cambria ETFs, the evaluation with begin in the first full calendar month since the fund’s inception and end in July 2016. Here is a chart with related statistics of the cumulative RealAlpha™ for the fund:
Similarly to its predecessors, the fund failed to outperform its reference ETF portfolio which had a slightly smaller volatility, measured as the standard deviation of monthly returns. The fund’s RealBeta™ was moderately higher than that of a broad-based domestic equity ETF.
The following chart and corresponding statistics show the constant composition of the reference ETF portfolio for the fund over the same period:
The fund had major equivalent positions in the Schwab International Small-Cap Equity ETF (SCHC), WisdomTree International SmallCap Dividend Fund (DLS), First Trust Dow Jones Global Select Dividend Index Fund (FGD), iShares MSCI United Kingdom ETF (EWU), PowerShares DWA Industrials Momentum Portfolio (PRN), and Vanguard FTSE Europe ETF (VGK). The Other component in the chart collectively represents additional five foreign-stock ETFs covering the New Zealand, Japan, Australia, Spain and Mexico markets. The reference weights indicate a significant foreign small-cap equity tilt of the fund.
Lastly, we will evaluate the Cambria Global Value ETF (GVAL). The issuer states that this product implements a proprietary index which
…consists of stocks with strong value characteristics. The Index begins with a universe of 45 countries located in developed and emerging markets. […] The Index next separates the top 25% of these countries as measured by Cambria’s proprietary long term valuation metrics. The Index then screens stocks with market capitalizations over $200 million. The Index is comprised of approximately 100 companies.
The following chart and associated statistics depict the cumulative RealAlpha™ for the fund:
Compared to its reference ETF portfolio, the fund added a modest amount of value (mostly in the last four months of the analysis period), although the portfolio had a slightly lower volatility. The RealBeta™ of the fund was substantially higher than that of a broad-based U.S. stock ETF.
The following chart and statistics demonstrate the fixed membership and weights of the reference ETF portfolio for the fund:
The fund had main equivalent positions in the iShares MSCI Italy Capped ETF (EWI), WisdomTree Europe SmallCap Dividend Fund (DFE), Guggenheim CurrencyShares® Euro Trust (FXE), iShares MSCI Poland Capped ETF (EPOL), iShares Latin America 40 ETF (ILF), and Global X MSCI Greece ETF (GREK). The remaining six ETFs in the above table, spanning the Spain, Brazil and Germany equities as well as international-corporate and emerging-markets bonds, collectively constitute the Other item in the above chart.
One of our previous posts outlined the benefits of similar analyses of iShares smart beta ETFs, which we will not repeat here for brevity. This evaluation of Cambria ETFs provides investors with similar insights.
Just like any other composite investment vehicles, Cambria ETFs change their holdings over time. Therefore, a question arises about the value of an analysis in which a static ETF portfolio is calculated from long-term data. The answer is to use a more advanced variant of Alpholio™ patented methodology, in which the membership of the reference ETF portfolio is still fixed but weights can fluctuate. Such a dynamic portfolio tends to more accurately track the analyzed fund over time.
For example, here is a chart with accompanying statistics of a reference ETF portfolio determined in that manner for the Cambria Shareholder Yield ETF (SYLD):
This gives a more accurate view of the fund’s recent average exposures.
If you would like to use our ETP Analysis Service to investigate similar products, please register on our website.
July 12, 2015
Inflation- and Dividend-Adjusted Market Peaks
The popular market-proxy S&P 500® index is market-cap weighted. This is one of the factors that helps reduce the turnover of ETFs tracking this index. For example, the iShares Core S&P 500 ETF (IVV) has a turnover rate of only 4%. The following chart, produced by the Alpholio™ App for Android, shows the characteristics of a portfolio composed solely of this ETF:
(Note that Alpholio™ uses a broader ETF as a representation of “the market”; hence, the beta of IVV is different from the conventional one and alpha from zero.)
However, market-cap weighting implies that the largest companies’ stocks have the highest impact on the index. While returns of mega-caps in the index tend to be less volatile, they are usually lower than those of their smaller-cap peers. To overcome this limitation, other ETFs weight equities in the index differently. For example, the Guggenheim S&P 500™ Equal Weight ETF (RSP) assigns each of the 500 stocks a 0.2% weight. This tilts RSP toward smaller-cap equities in the index and results in a 18% turnover. Over the same analysis period, RSP produced markedly higher returns than IVV but at the expense of an elevated volatility and a slightly lower Sharpe ratio:
In addition to overweighting of mega-caps, some economic sectors in the index dominate others, as shown in the latest edition of S&P Capital IQ The Outlook:
To counteract this, the ALPS Equal Sector Weight ETF (EQL) applies the same weight to nine sectors (with telecommunication services considered part of information technology). Here are the characteristics of a portfolio consisting solely of this ETF over the identical analysis period:
While the annualized return of EQL was lower than than of IVV or RSP, it was more than adequately offset by a decrease in volatility, which resulted in an improved Sharpe ratio and maximum drawdown.
What if the investor wanted to equal-weight all ten sectors instead of just nine, i.e. keep telecoms separate from IT? To do so, the investor could construct a portfolio of Vanguard sector ETFs, excluding the Vanguard REIT ETF (VNQ). That is because real estate stocks are currently part of the financials sector and not expected to become a separate asset class until mid-2016. Here is how such a portfolio, rebalanced quarterly (just like EQL), performed over the same analysis period:
The Vanguard sector portfolio had the second highest alpha and Sharpe ratio as well as the second lowest standard deviation (a measure of volatility of returns).
The above analysis period was dictated by the inception date of the EQL, the youngest of all the ETFs used. Arguably, this approximately six-year period may be considered too short and not representative of performance over a full economic cycle. However, it was interesting to see that while equal-weighting the index on a security level produced highest absolute returns, equal-weighting on a sector-level delivered the highest risk-adjusted returns.
To conduct your own analyses of various ETF portfolios, download the Alpholio™ app from
December 26, 2013
Year-End Market Predictions
With a strong performance of equities in 2013, major market indices have reached peak levels. In an attempt at accuracy, many articles in the financial media adjust index values for inflation. The conclusion is that the market, as measured by the Dow Jones Industrial Average (DJIA), has surpassed record levels from early 2000 only very recently on an inflation-adjusted basis. On the other hand, a broader market benchmark, the S&P 500® index, is still about 15% below its inflation-adjusted 2000 peak. Having suffered significant downturns in both 2000 and 2008, the NASDAQ-100 index is significantly below both its nominal and inflation-adjusted historical highs.
The problem is that in all of these assessments, a price level of each index is used. Would the findings be different if indices were also adjusted for reinvested dividends to account for total returns? To determine that, Alpholio™ compiled inflation- and dividend-adjusted prices of two representative exchange-traded funds: the SPDR® Dow Jones® Industrial Average ETF (DIA) and SPDR® S&P 500® ETF (SPY). These ETFs are long-lasting and popular implementations of their respective indices. To adjust for inflation, the Consumer Price Index – All Urban Consumers (CPI-U) was used.
A conventional price chart shows that DIA has indeed just matched an inflation-adjusted record high from January 2000:
However, a chart for DIA with reinvested dividends indicates that the previous inflation-adjusted peak from October 2007 was already surpassed in mid-January 2013:
Similarly, a price-only chart illustrates that SPY is still about 13% below its inflation-adjusted top from March 2000:
On the other hand, the chart with dividends factored in demonstrates that SPY already exceeded the previous inflation-adjusted maximum level in May 2013:
According to S&P Capital IQ, since the late 1920s dividends constituted about 45% of the total return of the stock market. Therefore, any assessment of the current market level has to adjust not only for inflation but also for reinvested dividends. From that standpoint, historical market peaks were quietly surpassed much earlier this year. Time and again, media focus is on generating simplified headlines rather than noting true events.
December 12, 2013
At year’s end, many analysts make market predictions for the next twelve months. The S&P 500® index is a popular target for such forecasts since it is commonly used as a market proxy and its constituent stocks are widely followed. Hence the bottom-up analysis — a sum of estimates for all individual equities makes an index forecast.
Who better to predict the S&P 500® index level than the S&P itself? Let’s take a closer look at their forecast accuracy. The following chart compares the predicted to actual values of the index, which Alpholio™ compiled from historical editions of S&P Capital IQ’s The Outlook:
To be exact, these 12-month targets were typically set in early to mid December of the preceding year, while the actual index values were recorded on the last trading day of the predicted year. Also, dividends were not taken into account in this price index.
The immediate takeaway from this chart is that the forecast for 2008 vastly overestimated the actual price: 1,650 vs. 903, or by about 83%! The financial crisis and its magnitude caught everyone, including members of the S&P Capital IQ’s Investment Policy Committee, by surprise. Excluding that outlier year, here are the index prediction and annual return statistics:
||Prediction vs. Actual
||Actual Index Return
The sample is admittedly small, under ten data points. But a trend is emerging — on average, S&P predictions underestimated the actual index. This tendency is further illustrated by the following chart from FactSet’s Targets & Ratings report:
The chart shows how a bottom-up target price (dashed line) moved almost in parallel with the actual index (solid line) in the 12 months through October. In other words, predictions were adjusted upwards with a lag as it became evident that original estimates were likely going to be soon surpassed. (As a side observation, almost half the stocks in the S&P 500® were rated a Buy, slightly less than half a Hold, and only about 5% a Sell. Even a booming market, a less optimistic distribution would be intuitively expected.)
S&P offers another interesting prediction for the next year:
We also believe 2014 could be one of those years in which the S&P 500 is up for the entire year but suffers through a pullback of 5%-10% (and more likely a correction of 10%-20%) before ending the year higher than where it started. One reason is that 26 months have elapsed without the S&P 500 slipping into a correction, versus the average of 18 months (and median of 12 months) between declines of 10% or more since 1945.
If statistically the market is overdue for a correction, let’s also hope that by the same token S&P underestimated the 500® index’s value in 12 months from now. In that case, we should be expecting a price of about 1930 instead of the current target of 1895, while keeping in mind that perfect market predictions are virtually impossible.