Tag Archives: automotive

Summary iShares factor ETFs provide a source of well tested algorithms for factor-based stock selection. Previous examination of QUAL, MTUM and USMV have shown that this approach can produce actionable investing ideas. Can adding other, well-documented, factors improve the selective powers of this approach?. I continue to think about mining the iShares smart beta ETFs for investing ideas. In this article, I want to discuss expanding the source of data to include ETFs for risk premium factors beyond those I looked at previously. Let me start by reviewing some recent results from this exercise. My starting premise is that the set of ETFs offered by Blackrock iShares emphasizing individual risk-premium factors provides a rich source of securities that have passed their quantitative filters for the target factors. Previously I looked at three of these ETF focused on low volatility, quality and momentum factors. My goal was to find stocks that appeared in the holdings from more than one of these ETFs with the idea that such stocks have passed the MSCI index screen for more than one factor. I identified 14 stocks that occur in all three ( A Quest for the Smartest Beta ) and 60 more that occur in at least two ( Can We Find Smarter Beta From 2 Factor Portfolios? ). I found the results intriguing. First, The ETFs all beat the market, as represented by the SPDR S&P 500 Trust ETF ( SPY), as does the equal-weighted portfolio of ETFs. By analyzing a hypothetical portfolio, I was able to show that the 14 holdings from the set occurring in all three ETFs has soundly beaten all of the ETFs as well as the equal-weighted portfolio of the ETFs. This is fully documented in the second article referenced in the previous paragraph. Readers commented on my omission of two of the classic risk-premium factors and offered suggestions on incorporating them into the models. The missing factors, value and size, are, of course, important, and I’m going to look at how much, if anything, they add to the exercise as I go on. But first, let me digress here for a paragraph or two and consider why I felt these factors could, or should, be left out. Let’s start with the objective: It is to mine the quantitative algorithms of MSCI’s factor indexes for high-potential stocks. As I explained in the second article, I wanted to keep this exercise to a manageable number of funds and holdings. I thought three was optimal. Also, value and size are much less straightforward to deal with in this context. These factors form the basis for the traditional classifications of stocks: Value vs Growth and Large-, Mid-, Small-Cap. It’s the Morningstar style box. Value is variously defined and it’s not at all unusual to see the same securities turning up in growth and value funds from the same group. Size is easy, but pairs poorly with other factors depending on how one makes size cuts. By contrast, quality, momentum and low volatility are less rigorously defined (even considering the vagueness of how value is defined) and, in my view, more amenable to quantitative analysis that can produce unique, actionable results. So, I went with quality, momentum and low volatility. Quality is something I’ve been thinking about a lot, and I like the algorithm QUAL is using to define the factor (discussed here ). Momentum is another factor that can add serious alpha. I’ve been maintaining some momentum-based investing strategies in moderate-size portfolios using commission-free ETFs for several years to modest success. A problem with momentum is it tend to generate volatility and I’ve tried to modulate that in my own investing by adding a weighting for volatility (some day I may write an article on this). This reflects my appreciation for low volatility and the thinking that led me to include USMV in this project. The Factor ETFs All this is a bit subjective and intuitive, which is always something to guard against in an evidence-based approach, so I’ve decided to take readers’ advice and look at two more of iShares MSCI factor-index funds. I wanted to see if adding value and size to the analyses can improve the results. To this end, I’ll be deconstructing five ETFs looking for common holdings. The list of five, starting with the three considered earlier: iShares MSCI USA Minimum Volatility ETF (NYSEARCA: USMV ), iShares MSCI USA Momentum Factor ETF (NYSEARCA: MTUM ) iShares MSCI USA Quality Factor ETF (NYSEARCA: QUAL ) iShares MSCI USA Value Factor ETF (NYSEARCA: VLUE ) iShares MSCI USA Size Factor ETF (NYSEARCA: SIZE ) One problem right off the bat is the size of SIZE. At 636 holdings, it’s nearly four times the size of the next largest fund (USMV with 165). Perhaps as a consequence, it adds little value to the analysis, although, despite having 636 holdings, it is the least correlated with the broader market of the five ETFs. (click to enlarge) Pay particular attention to the last column in that table. SIZE is the least correlated with SPY, much lower than I would have anticipated. Note too, that VLUE is less correlated with SPY than any of the other three ETFs. I’ll start by looking at the performance of these ETFs and ask if the two new additions look likely to add any value. (click to enlarge) For the past year, they have lagged the previously considered three. But this has not been a good year for value stocks, and SIZE may add an advantage from that low correlation coefficient that will only become evident when it becomes an important variable. Deconstruction the ETF Portfolios As I did previously, I downloaded the full holdings of each of the ETFs into a spreadsheet and analyzed all five for stocks that appeared in more than one of the funds. Here’s a summary of the results. As anticipated, it quickly gets unwieldy. Only a single stock is in all five funds, and there are 20 that appear in four of the ETFs. Beyond that, there are too many to be useful for my purposes. What’s interesting is the 14 stocks that formed the basis of the earlier analysis by occurring the holdings of QUAL, MTUM and USMV, are all included in the 21 four- or five-fund stocks here. Thirteen of the 14 occur in either VLUE or SIZE; only one is in both. So, if we take the top 22 stocks here, i.e. those occurring in at least four funds, we have added eight to the previous list. So far, so good, we have increase our candidate pool; but not excessively, it’s still a manageable number. Here, for the record, are the 22 stocks with the 14 from MQLV set in italics: Axis Capital Holdings Ltd (NYSE: AXS ), Accenture Plc (NYSE: ACN ), Ace Ltd (NYSE: ACE ), Arch Capital Group Ltd (NASDAQ: ACGL ), Assurant Inc (NYSE: AIZ ), AT&T Inc (NYSE: T ), Chevron Corp (NYSE: CVX ), Chipotle Mexican Grill Inc (NYSE: CMG ), Chubb Corp (NYSE: CB ), Eli Lilly (NYSE: LLY ), Home Depot Inc (NYSE: HD ), Nike Inc Class B (NYSE: NKE ), O’Reilly Automotive Inc (NASDAQ: ORLY ), Partnerre Ltd (NYSE: PRE ), Reynolds American Inc (NYSE: RAI ), Sigma Aldrich Corp (NASDAQ: SIAL ), Starbucks Corp (NASDAQ: SBUX ), Target Corp (NYSE: TGT ), Travelers Companies Inc (NYSE: TRV ), United Health Group Inc (NYSE: UNH ), Visa Inc Class A (NYSE: V ), WR Berkley Corp (NYSE: WRB ). The first entry, Axis Capital, is the single name in all five ETFs. Sector representation is dominated by Financials and Consumer Discretionary, but it is more diverse than the set of 14 derived from three ETFs. (click to enlarge) Here is how these 22 stocks are allocated among the ETFs. As we see, all are in SIZE. SIZE is therefore acting as a binary filter to select among funds that are in three of the four funds but do not pass the size-factor filter. This is potentially a useful filter. USMV holds all but one, so it’s a similar filter. VLUE is a stronger filter. Less than half the funds are in VLUE’s holdings. I find this interesting and would have expected a result like this from MTUM, which only misses four, none of which is likely to be mistaken for a momentum stock in the current market. As I refine my thinking on this whole exercise, I have to spend more time considering how VLUE affects results. Portfolio Analysis As previously, I wanted to see the results of this filtering process. There is only one record to analyze. The funds rebalance at the end of May and November and, to my knowledge, do not publish past index allocations. Thus, there is only one analyzable record, that for the current cycle which is about 5 months old. We can see how various permutations of these results have fared since the last rebalance. I ran analyses on Portfolio Visualizer for equal-weighted portfolios comprising the following with the coding I’ve used in the tables: Five ETFs: 5ETFs EW QUAL, USMV, MTUM: 3ETFs (QVM) EW Stocks present in holdings of at least 4 of the ETFs: VQMVS(4+) VQMVS(4+) stocks in QUAL and MTUM only: QxM VQMVS(4+) stocks in QUAL and VLUE only: QxV VQMVS(4+) stocks in MTUM and VLUE only: MxV I pulled out the last three sets because USMV and SIZE were doing little more than serving as a final filter for the other three ETF holdings’ overlaps, so I thought it useful to see how those components were contributing to the results. Here are those results. (click to enlarge) As we can see, the five ETFs as an equal-weighted portfolio beat SPY, but lagged the subset of three ETFs. Let’s not forget, however, that this is only a five-month result. Longer term results can show benefit to holding all five factor ETFs, or at least four of them. For this we do have a longer record. The full record is still limited as the youngest fund only dates to July 2013. From July 2013, equal-weighted portfolios, rebalanced semiannually, of combinations of five, four and three of the ETFs turned in the following performance results. (click to enlarge) Removing either SIZE or VLUE added return and reduced maximum drawdown. Removing both, i.e. going to only QUAL, MTUM and USMV, as previously considered, improved both metrics. Volatility did increase slightly, but in all cases it remained lower than the S&P 500. These results indicate that there has been no advantage to adding VLUE or SIZE to a factor-based ETF portfolio. I’d like to say this validates my decision to use only MTUM, QUAL and USMV in my analyses, but the fact remains that the data set is too limited to draw such a conclusion. Let’s return to the previous table – and our main topic – and see how stocks filtered from the ETFs on the basis of their presence in four or more funds fared. Over the past five months, the combined ETFs returned 1.40% CAGR for all five, and 5.75% CAGR for the MQLV three. A portfolio of the 22 stocks found in four or more ETFS 29.67% CAGR and did so with a max drawdown of only -3.35% vs. -6.52% for the better performing of the two ETF portfolios. Separating out the component ETFs we see that the combination of QUAL an MTUM added a remarkable level of value, far outpacing a combination of either of the two factors with value as represented by VLUE. Yet again, I must emphasize the limited data available. But the results certainly begin to suggest that these ETFs, especially MTUM, QUAL and USMV, are attractive sources for filtered lists of stocks that rank strongly for risk-premium factors which can be further filtered for having been selected by the quite different quantitative criteria by multiple funds.