“It’s almost getting too scientific where every sector’s going to have this exact exposure. We’re going to have these ten longs and ten shorts. We’re going to rip out every factor element of it. That’s what I call over-engineering. It seems like the benefit versus brain damage is not a good quotient.”
-Wes Gray
Today’s quote is pulled from the founder of Alpha Architect in one of the better discussions I’ve come across on quantitative investing from Investor’s Field Guide (LINK). Alpha Architect has created several ETFs with a quantitatively-driven process to isolate pure factors. The goal is to be concentrated in the factor each ETF is claiming to capture.
As can likely be gleaned from the quote, the belief is that some strategies sway from the concentration and pure disciplined factor exposure needed to capture the said goal (value, momentum, growth, etc.). Any disconnect probably depends whether or not the marketing pitch fits the results.
The sophistication level and actual factor mechanics within the popular factor buzzwords clearly varies, but we do feel comfortable saying that automated strategies and execution from a larger percentage of participants who seek to capture factor elements has changed broader market structure rapidly in the last decade…
On our team we find the evolution of market structure and investor motives fascinating and embrace the urgency to stay on top of it. There is a heavily-weighted behavioral component to our process because we have the tactical overlay to our macro research.
Back to the Global Macro Grind…
When we think about asset allocation and exposures, we think about it through the lens of growth and inflation. Each series is either accelerating or decelerating within a positive or negative slope.
This far in advance, we’ll call it an empirically supported estimate, but our GIP model has the U.S. economy tracking in a QUAD 1 environment for each of the next 3 quarters (growth accelerating as inflation decelerates). Beginning in Q3 our YY GDP estimates become widely divergent from consensus Wall Street estimates (the consensus estimates have nothing to do with the math behind QUAD1):
- Q3 2017: +2.59% YY vs. +2.10% consensus
- Q4 2017: +2.79% YY vs. +2.20% consensus
- Q1 2018: +3.19% YY vs. +2.50% consensus
Retail Sales reported yesterday morning for the month of July provided an expected and supportive boost to this QUAD1 reality. Retail sales is a major contributor to real growth for any given quarter in our GIP model:
- Headline retail sales accelerated to +4.2% YY vs. +3.4% YY in June
- The retail sales control group (the series that is represented in GDP) that feeds into our GIP model accelerated to +3.6% YY in July from +2.5% YY in June. The Y/Y delta in the control group is the highest since March
Within our tested framework, a QUAD 1 environment has historically been:
- Good for equities in general
- Even better for what we would call the “Real Growth” exposures within equities like Technology, Consumer Discretionary and the Russell 2000, all relative to the S&P 500
At this point, the discussion between our preferred “real growth” exposures and traditional quantitative factors like growth, value, and momentum probably diverge. The exposures listed above are our preferred real growth exposures because they tend to outperform in QUAD 1 when real growth is accelerating. The important difference is that we are not referring to most definitions of “growth” as a quant factor although clearly there is overlap. Our real growth exposures consist of a number of different “factors” in the traditional sense.
Back to economic growth, assuming the economy does accelerate like our GIP model suggests, what does this incremental growth mean for Factor Exposure (i.e. Russell 2000 Growth (IWO) vs. Value (IWN) with the Russell 2000 index)?
There is unquestionably a momentum of sorts in those two factors right now, and getting in front of “exhausted momentum” before the broader rotation turns is a quant golden goose. Will there be an IWO to IWN rotation if the economy does really accelerate as a credible argument is that investors hide out in some vehicles that fit the “growth” label when growth is actually scarce (this could me more of a capitalization factor argument)? Or has the market already front-run real growth accelerating in what may be a QUAD 1 environment for three quarters? It’s an important question when comparing some of the sector and factor divergences YTD:
- IWO +7.8% vs. IWN -3.3%
- XLK +20.1% vs. XLE -15.7%
Again sticking with the IWO to IWN debate, either one of those two baskets of stocks will have momentum at any given time, and most of the IWO basket also has the momentum factor on its side in 2017.
Without a process-driven reason for us to get off the QUAD 1 exposure train despite the fact that it’s been the right train for multiple months, our quantitatively-generated volatility factors have been vital to risk managing all-time highs in 2017.
Our real preferred real growth exposures have hit multiple new all-time highs, outperformed, and also screen in the “momentum” category. This has led to an unprecedented volatility smash followed by short-winded bouts of computer-driven turbulence, volatility, and a lot of anxiety about a broader correction.
So although a portfolio more heavily weighted toward the Russell 2000 growth ETF has outperformed the Russell 2000 ETF benchmark by ~600bps in 2017, the sequencing in volatility and volatility expectations have suggested on multiple occasions, that “today is not the day” to get more heavily exposed to this factor within the benchmark.
To explicitly provide an example of the risk associated with the IWO vs. IWN momentum, we’ll reference the Chart of the Day which is a ratio series of at-the-money front month implied volatility in the Russell 2000 value ETF relative to the Russell 2000 growth ETF. Due to the fact that IWO has outperformed on any duration specified over the last 12 months, implied volatility still comes at a premium in IWN despite the fact that IWN is historically a lower vol exposure. And this ratio series typically spends most of its time well below zero.
We’ll take a snipit from our daily morning note from Friday, August 4th to explain our process because there is no denying the performance and psychological momentum in the growth component of the Russell 2000 (the Russell was just an example to hash out the factor point and proprietary risk management process):
|
“From Tuesday of this week: ‘For the last couple months we’ve hit on this divergence in Russell 2000 expectations relative to the Nasdaq. Both tend to do well in QUAD 1, but even just last week, we were looking at all-time lows in implied volatility expectations across durations in the Russell 2000 and its different factor components which has clearly been quite a bit different from the Nasdaq and/or the QQQs.’ Often, the vol smash bottoms out with the high-end of the risk range. In a low vol environment in something we like, straddle or strangle buying is attractive at the high end of the risk range when implied volatility pancakes. This is really just a long vol, directional hedge (long a call and a put near the money with the logic of booking the gain in the put on a pullback and keeping upside exposure in something you like at a better price). If there ever was an environment to use this strategy, the current environment looks attractive in the Russell 2000 in my opinion – clearly not after the last couple of days.” |
Our immediate-term Global Macro Risk Ranges (with intermediate-term TREND views in brackets) are now:
UST 10yr Yield 2.19-2.30% (neutral)
SPX 2 (bullish)
RUT 1 (neutral)
NASDAQ 6 (bullish)
XOP 29.64-31.43 (bearish)
DAX 115 (bearish)
VIX 9.31-16.17 (bearish)
USD 92.51-94.00 (neutral)
Oil (WTI) 47.12-50.06 (bearish)
Good Luck Out There Today,
Ben Ryan
