Why lawyers seem so long-winded
Have you ever looked with glazed eyes at legal documents and wondered why it seems that lawyers seem so determined to say things in such twisted ways? Why can't they just be straight-forward, like everyone else? The answer is that they need to be capable of responding when prospective clients like PowerShares show up at their offices.
In the last blog, we saw how ETFs may use the language of passivity in their formal documents ("ABC ETF aims to track as closely as possible the dally performance of the XYZ Equity Index") but offer ways to implement highly active strategies (i.e. the my-index-is-better-than-yours approach).
Thus far, the dueling indexes have pretty much been name-brand benchmarks put out by the likes of S&P, Dow Jones, Wilshire, etc. They strayed a bit beyond the notion of being broad market barometers by offering value-only and growth-only variations, or different flavors based on market cap. Such variations have been enough to incite objections by commentators who still think there is a one inherently-correct, good-for-all-time passive strategy, but not enough to attract much attention from serious quants.
That was then. Now, we have a new class of ETFs that loudly trumpet their desire to outperform the traditional market benchmarks. In other words, we have ETFs that passively track some indexes that are designed to outperform other indexes. Sound confusing? Don't worry about it. Leave it to the lawyers. Suffice it to say that the "my index is better than your index" theme has evolved to become: "my index, which you never heard of because it was created solely for the purpose of being used by one specific ETF, is way better than all the generic indexes you and everybody else have been looking at for years."
If we can recognize that Alpha is generally computed with respect to a recognized benchmark like the S&P 500 and look the other way when lawyers toss the language of passivity into the prospectuses, we can recognize PowerShares XTFs and competing products as alpha-seeking indexes-ETFs.
How alpha-seeking indexes-ETFs work
The best way to get a sense of how all this works is to look directly into a PowerShares prospectus, specifically for this example, the one for the PowerShares Dynamic Aggressive Growth Portfolio (PGZ). Here is how the ETF's principal investment strategies are described.
The Fund will normally invest at least 80% of its total assets in common stocks
of aggressive growth companies. The Fund will normally invest at least 90% of
its total assets in common stocks that comprise the Aggressive Growth Intellidex.
The Aggressive Growth Intellidex is comprised of 100 U.S. aggressive growth
stocks selected on the basis of their capital appreciation potential as identified by
the AMEX (the "Intellidex Provider") pursuant to a proprietary Intellidex
methodology. A stock's characterization as "aggressive growth" is based on a
multi-factor methodology designed to create distinctions between stocks that
technically qualify as growth stocks and stocks that exhibit strong growth
characteristics. This style delineation process incorporates a unique blend of
factors that rely on accounting and consensus analyst data and the combined
estimates of analysts that cover such stocks, and forecasted earnings in the style
to create distinct style groups within the growth classification. Once a stock's
growth characteristic is determined, stocks that do not exhibit strong growth
characteristics are excluded.
Here is how the Intellidex methodology is described.
The Aggressive Growth Intellidex methodology is designed to objectively identify
those stocks within the aggressive growth market segment that have the greatest
potential for capital appreciation. The methodology evaluates companies
quarterly, based on a variety of criteria, including fundamental growth, stock
valuation, investments and risk factors, and then ranks and sorts them based on
their cumulative scores. Component stocks for the Aggressive Growth Intellidex
are selected from among the companies with the highest-ranking cumulative
score ("Model Score") within an aggressive growth universe.
A familiar ring
Portfolio123 users should recognize this process. It's what many have been doing all along with the Stock screener, ranks and the automated portfolios that are established after working with simulation.
The specific index construction protocol used by Intellidex is a bit more formal.
(1) The 2,000 largest U.S. stocks (by market capitalization) traded on the New
York Stock Exchange ("NYSE"), the AMEX and the NASDAQ are ranked for
capital appreciation potential using a proprietary AMEX Intellidex model.
(2) The universe of companies is divided into groups based in the following manner:
(a) The universe of stocks is segregated into three style groups: aggressive
growth, value or core. Core includes stocks which technically may be
considered growth stocks but which do not exhibit strong growth
characteristics. A stock's characterization as aggressive growth, value or
core is based on a multi-factor methodology whose measurements
include forecasted price to earnings, price to book, dividend yield,
forecasted earnings growth, revenue and book value core. 500 of the
stocks in the total universe are deemed aggressive growth, 500 of the
stocks in the total universe are deemed value and 1,000 of the stocks in
the total universe are deemed core. Core and value stocks are excluded
from the Aggressive Growth Intellidex.
(b) The aggressive growth universe is ranked by Model Score. The 100 top
ranked stocks from the aggressive growth universe are selected for
inclusion in the Aggressive Growth Intellidex and equally weighted to one
percent each.
But the general idea is the same. It's impossible to determine if Portfolio123 was actually used in any part of the research process. But based on the general outline presented here, it's at least possible.
These PowerShares ETFs, and others from other families that seek to follow this approach, are grouped together by the Custom ETF Universe creation interface under the Method known as "Special Models." You can also spotlight this group in the screen via ETFMethod=QUANT.
Good news . . . so-so news . . . hope that we can get back to good news
The good news lies in the fact that this category of ETF holds so much potential. We at Portfolio123 know full well how rewarding rules based strategies can be. It has to feel satisfying to see that such approaches are now made available as ETFs.
The so-so news, at least so far, has been in the extent to which such ETFs have actually succeeded.
Here are the rules for a simple screen I created to identify these ETFs.
ETFCountry=USA
ETFAssetClass=EQUITY
ETFMethod=QUANT
Table 1 shows the result of an advanced ETF backtest I conducted assuming a new portfolio every week, a four-week holding period for each portfolio, and a start date of 3/3/05, which is when the number of these ETFs first reached double digits.
Table 1
|
Average 4-week % change |
| ETF Portfolio |
S&P 500 |
Excess |
| Overall |
-0.43% |
-0.65% |
0.22% |
| Up Markets |
2.85% |
2.31% |
0.54% |
| Down Markets |
-4.62% |
-4.44% |
-0.18% |
It's not a disaster. Overall, the group was a bit better than the S&P 500. So in a very strict sense, I suppose one could say this category is succeeding. But realistically, the degree of success is small (too small, it seems, to accommodate slippage), and conventional backtests will vary depending on when you begin and end the test periods. And clearly, this is not enough to justify the sense of excitement generated by the marketing literature. (The results do not meaningfully change if I add the rule Enfamil=POWERSH to limit consideration only to PowerShares, the leading brand in alpha-seeking ETFs.)
But looking ahead, there is reason to hope for better things from this group. Bear in mind it didn't get up and running until 2005. In the spring of 2006, the market experienced a hit which resulted in the sort of factor breakdowns and reversals we would see much more of as we moved on and especially in 2008, which, as we know all too well, was a disaster. It's hard to criticize these ETFs for pedestrian performance when so many models created by so many investors completely imploded. (Things were so bad for quant models, PowerShares et. al. might even have the right to brag about having been able to more of less match the S&P 500).
Table 2 shows the results of an advanced backtest with successive weekly PowerShares-only portfolios formed from 4/2/05 through 4/1/06.
Table 2
|
Average 4-week % change |
| ETF Portfolio |
S&P 500 |
Excess |
| Overall |
1.64% |
0.82% |
0.82% |
| Up Markets |
2.72% |
1.72% |
1.00% |
| Down Markets |
-1.07% |
-1.45% |
0.38% |
That's more reasonable. It's not spectacular, by any means. But it looks like this sub-universe may have some potential for later on, when the markets normalize.
When I add a rule that eliminates sector-specific quant funds (ETFSector=GENSECT), the result is similar: Excess portfolio 4-week performance overall, in up months and in down months, averages 0.77%, 0.95% and 0.32% respectively.
Thinking about a more usable strategy
Based on these results, I don't have high hopes for immediate implementation of quant-ETF strategy since the group as a whole is cold. But given that there does seem to be some indication of potential for when the market is no longer experiencing the kind of liquidations we've been seeing, I decided to start thinking out loud about how one might later choose among quant ETFs.
I started with a basic screen to identify PowerShares domestic-equity quant ETFs that are not sector-specific.
ETFCountry=USA
ETFAssetClass=EQUITY
ETFMethod=QUANT
ETFFamily=POWERSH
ETFSector=GENSECT
I eliminate sector ETFs in this context since their performance may be more heavily influenced by what's happening in the sector, rather than by the overall quant model. Sector ETFs, quant and otherwise, seem more amenable to a rotation-seeking approach.
We need to be careful here about aggressive filtering once we've defined this universe because it's very small, only 22 ETFs as of this writing.
Considering what these funds claim they want to do, it would seem reasonable to consider a rule based on strong fund performance relative to the S&P 500. So assuming that's the benchmark to be used by the screen, I add one more rule:
Close(0)/Close(60)>Close(0,#BENCH)/Close(60,#BENCH)
Figure 1 shows the result of a backtest run from 3/31/05 through 3/31/06 assuming 4-week rebalancing, 0.5% slippage and no limit on the number of ETFs (which happened to always be in single-digits back then).
Figure 1
If I change the screen to work with share price performance over the past 20 trading days, it does not backtests well; it just keeps pace, more or less, with the S&P 500. But if I switch the measuring period to 120 trading days, performance pretty-much matches the 60-day version of the rule.
This indicates that monthly performance (which is generally what we have with 20 trading days) is almost random in terms of whether or not a successful quant fund will perform as expected. But when we stretch our measuring period further out, the better models seem to have been more likely to stay better.
Meanwhile, Figure 2 shows how the screen (using the basic 60-day measuring period) performed during the "bad" period, 4/1/06 through 2/28/09.
Figure 1
That's certainly nothing to brag about. But as many of know, the period was terrible for just about all sorts of quant modeling approaches, and it worsened dramatically as it progressed. All things considering, a portfolio of quant ETFs that matches the S&P 500, after allowing for .5% slippage, is really not the end of the world.
This isn't to say we'd want to implement a quant-ETF strategy like this right now. But it does seem reasonable to say it may be productive to start developing strategies now, working with a backtest period of 3/31/05 thru 3/31/06, for further refinement and potential application in the future, when market conditions improve and when alpha-seeking ETFs hopefully deliver on their expectations.
This isn't just an ideal hope for the future, as it might be if we were to assume banks will recover to their old highs or that GM and Ford will regain old glories. Portfolo123 users who are experienced with quantitative rules-driven equity strategies know more deeply what these ETFs are trying to accomplish, understand what went wrong of late, and should be able to recognize when conditions for this approach improve simply by monitoring their own models.
Next steps
The next blog in this series will move to a very closely related area, fundamental weighting, which bears an interesting theory, an impressive intellectual pedigree (Jeremy Siegel, Robert Arnott), lots of great press (at least until recently), and some very challenging performance issues. Then, we'll be ready to tackle ProShares, Rydex, and Direxion, ETFs that are about as aggressive and interesting as any funds can get.
