Our research has an explicit focus on regime segmentation to reduce the range of potential economic and financial market outcomes to a narrow sample that is more probable and, thus, more manageable for investors. We accomplish this goal by first identifying four distinct states an economy can operate in from the perspective of the trending Rate of Change of economic growth and inflation – the first and second principal components for the preponderance of asset class returns:
1. GOLDILOCKS: GROWTH momentum ↑ and INFLATION momentum ↓
2. REFLATION: GROWTH momentum ↑ and INFLATION momentum ↑
3. INFLATION: GROWTH momentum ↓ and INFLATION momentum ↑
4. DEFLATION: GROWTH momentum ↓ and INFLATION momentum ↓
From there we employ a combination of proprietary econometric tools – namely our Stationary Mean Reversion Model and our Agent-Based Nowcast Model – to project the most probable path for GROWTH and INFLATION over our NTM forecast horizon. We use those monthly GROWTH and INFLATION projections to derive GRID outcomes for each month, which we denote as the bottom-up Macro Regime – i.e. what the economy is doing in trending Rate of Change terms.
We have also carefully backtested every major financial market instrument, across asset classes, from the perspective of our historical bottom-up Macro Regime observations. Those backtests allow us to understand the general behavior of each exposure within and across the four GRID Regimes through the quantitative lenses of annualized returns, percent positive ratios, volatility, and covariance. That data is programmed into our Global Macro Risk Matrix, which scores the Volatility-Adjusted Momentum Signals of 42 of the most liquid and widely followed asset markets in accordance with those findings. At every interval, any combination of asset markets could be independently confirming of each GRID Regime – and to varying degrees over time. This repeatable confirmation process helps us quickly and cogently identify how asset markets are generally behaving on a trending basis, which we denote as the top-down Market Regime.
Our Volatility-Adjusted Momentum Signal (VAMS) is a dimension-reduced dynamic factor model based on a cumulative scoring system whereby an asset receives -2pts (bearish), 0pts (neutral), +2pts (bullish) via a -1pt/+1pt contribution if the local price regime is below/above its medium-term price regime and via a -1pt/+1pt contribution if local volatility regime is above/below its medium-term volatility regime – or vice versa, depending on the asset’s historical correlation to its implied or realized volatility. For example, Bitcoin’s price is positively correlated to its realized volatility, whereas the S&P 500 and the VIX tend to be inversely correlated.
Our Probable Ranges are quantitative risk management signals designed to spot inflection points in the short-term momentum of a security or asset class. Our system offers a thoughtful evolution to Benoit Mandelbrot’s Rescaled Range framework by explicitly linking the calculus of the Upper and Lower Boundaries of the Probable Range to a given exposure’s VAMS reading, whereby the range is asymmetrically skewed to the upside if the exposure is bullish VAMS and asymmetrically skewed to the downside if the exposure is bearish VAMS. This process helps investors reorient their risk management approach – i.e. buying dips vs. selling rips – much faster than traditional trading ranges in/around inflections in the medium-term price momentum and volatility characteristics of the asset.
Our Market Regime nowcasting process is a three-step, iterative procedure that we repeat daily:
1. Each bullish VAMS ✔ or bearish VAMS ❌ reading in our Global Macro Risk Matrix adds +1pt to each GRID Regime the signal makes sense in from the perspective of our GRID Asset Market Backtests. For example, a bullish VAMS ✔ in the S&P 500 would add +1pt to GOLDILOCKS and +1pt to REFLATION – both “risk on” regimes in our GRID framework. Conversely, a bearish VAMS ❌ reading in the S&P 500 would add +1pt to INFLATION and +1pt to DEFLATION – both “risk off” regimes in our framework. Neutral VAMS ! do not contribute to the scores of either GRID Regime.
2. That process is iterated across each macro market indicator in our Global Macro Risk Matrix to generate scores for each GRID Regime for each trading day. The cumulative score across the four GRID Regimes = Total Confirming Asset Market VAMS. A GRID Regime’s share of the total is equivalent to its Market Regime Probability. At every interval, the GRID Regime that earns the modal outcome distinction is said to be the Dominant Market Regime.
3. On average, there are about 2.5 material Dominant Market Regime phase transitions each year. Material = going from a “risk on” regime (i.e. GOLDILOCKS or REFLATION) to a “risk off” regime (i.e. DEFLATION or INFLATION) – or vice versa – each year. Lateral pivots are far easier to risk manage. For example, being long or short duration is the only toggle of substance between GOLDILOCKS and REFLATION, while being long the dollar and short foreign currencies, energy, food, and breakevens – or vice versa – is the only toggle of substance between DEFLATION and INFLATION.
Our primary mission is to help the investors on #team42 compound returns over time by first avoiding major drawdowns in their portfolio(s) and secondarily being long of assets that are likely to outperform according to our bottom-up Macro Regime projections. We accomplish this objective primarily by orienting our risk management – i.e. WHAT are we buying and selling; WHEN are we buying or selling; and HOW MUCH cash are we raising or deploying – in accordance with the Dominant Market Regime being nowcasted by our Global Macro Risk Matrix, which has generated clear “risk off” signals by transitioning to INFLATION or DEFLATION ahead of each of the major drawdowns in the S&P 500 since the start of 1998 on an out-of-sample basis:
- Summer 1998: our Global Macro Risk Matrix transitioned to DEFLATION on 6/3/98, roughly six weeks ahead of a -19% decline in the S&P 500
- 2000-02: our Global Macro Risk Matrix transitioned to INFLATION on 3/27/00, right at the outset of an eventual -49% decline in the S&P 500
- 2H07: our Global Macro Risk Matrix transitioned to DEFLATION on 8/1/07, well ahead of an eventual -57% decline in the S&P 500
- 2H08-1Q09: our Global Macro Risk Matrix transitioned to DEFLATION on 8/5/08, two months ahead of the height of the Global Financial Crisis
- Spring/Summer 2010: our Global Macro Risk Matrix transitioned to DEFLATION on 5/11/10, less than two weeks into an eventual -16% decline in the S&P 500
- Summer/Fall 2011: our Global Macro Risk Matrix transitioned to DEFLATION on 7/19/11, nearly three weeks ahead of a -19% decline in the S&P 500
- August 2015: our Global Macro Risk Matrix transitioned to INFLATION on 7/7/15, two weeks ahead of a -12% decline in the S&P 500
- Winter 2015-16: our Global Macro Risk Matrix transitioned to DEFLATION on 12/14/15, two weeks ahead of a -14% decline in the S&P 500
- Q4 2018: our Global Macro Risk Matrix transitioned to DEFLATION on 8/20/18, one month ahead of an eventual -20% decline in the S&P 500
- Q1 2020: our Global Macro Risk Matrix transitioned to DEFLATION on 2/4/20, two weeks ahead of a -34% decline in the S&P 500
Another way we attack the problem set that is quantifying how much downside risk exists in risk assets over the near-term is via monitoring the combined Market Regime Probabilities of INFLATION and DEFLATION – the two “risk off” GRID Regimes in our Market Regime Nowcasting Process – through the lens of our Cross-Asset Correction Risk Indicator, or “CACRI” for short. The median CACRI reading for the 12 eight-plus percent S&P 500 corrections in our post-GFC sample is 15% with an interquartile range of 900bps. As such, investors should be concerned if CACRI falls to the low-20s and they should be actively booking gains in risk assets if CACRI breaks down into the teens. From a flows perspective, we interpret a teens reading in CACRI as a signal that the momentum strategies employed by CTAs, Quants, and Macro Funds are at/near peak leverage.
- Q1 2010 -8%: CACRI bottomed at 14% on 1/12/10
- Summer 2010 -16%: CACRI bottomed at 16% on 4/13/10
- Summer/Fall 2011 -20%: CACRI bottomed at 11% on 4/12/11
- Spring 2012 -10%: CACRI bottomed at 8% on 3/23/12
- Fall 2012 -8%: CACRI bottomed at 16% on 9/28/12
- August 2015 -12%: CACRI bottomed at 14% on 5/7/15
- Winter 2015-16 -14%: CACRI bottomed at 37% on 12/2/15
- Volmageddon -10%: CACRI bottomed at 11% on 1/12/18
- Q4 2018 -20%: CACRI bottomed at 16% on 9/28/18
- COVID Crash -34%: CACRI bottomed at 10% on 1/3/20
- September 2020 -9%: CACRI bottomed at 16% on 8/26/20
- October 2020 -9%: CACRI bottomed at 21% on 10/16/20
We measure and map investor CROWDING via the percentage spread between 30-day at-the-money put implied volatility and the local realized volatility regime – which we estimate via the mean of 10-day, 20-day, and 30-day Rvol – with the following working hypotheses in mind:
1. Wide Ivol PREMIUM on elevated Rvol is indicative of crowding on the short side of the trade, which implies the exposure is due for a bounce;
2. Wide Ivol DISCOUNT on depressed Rvol is indicative of crowding on long side of the trade, which implies the exposure is due for a correction; and
3. Wide Ivol PREMIUM/DISCOUNT on depressed/elevated Rvol is indicative of volatility doing what it does best – i.e. mean revert – and while it is not especially actionable in the moment, it tends to have forward-looking implications – i.e. “climbing a wall of worry”/“bottoming process”.
We consistently study MoM Sharpe Ratio DISPERSION among 50 US equity sectors and style factors for three primary reasons:
1. It gives us a general sense of what GRID Regime other investors implicitly or explicitly think we are in;
2. It affords us a view on what sectors and style factors “Pod Shops” (i.e. multi-manager hedge funds) are forcing their PMs into and out of either implicitly through delta hedging activity or explicitly through capital reallocation; and
3. It affords us a lagging view on what sectors and style factor exposures “Pure Quants” are bidding up and selling down.
