It's weeks like these that remind me of the concept of Gell-Mann amnesia. Briefly stated, the effect occurs when we read an article in our own field of expertise and recognize the significant number of errors or over-simplifications made by the journalist. However, when we read an article outside of our field, we assume it is accurate.
If I can give one piece of advice to anyone in times like these, it is to go read a few articles written by journalists about a field you're an expert in and highlight all the things that are wrong. Then go read a few articles about the markets this week. Or politics. Or the Coronavirus. Then assume there are just as many mistakes made; you just don't know where.
If we recall past sell-offs, we'll find that there are no shortage of explanations and theories. Remember the 2010 Flash Crash? Theories included a fat-finger trade, the impact of HFT on market micro-structure, large directional bets, and technical glitches.
Regulators ultimately concluded it was all caused by a mixture of fragmented and illiquid markets and an unusually large and sudden sale of S&P 500 E-Mini futures contracts. The execution algorithm selected by the trader gave no regard to price changes, only prior volume, ultimately setting off a chain reaction among market makers that caused the market to plummet.
Technically interesting. Not a great headline.
In the remainder of this note, I want to spend some time walking through some broad market thoughts, what we are seeing in our quantitative signals and what it means for future positioning in our tactical portfolios.
But before we dive in, let's start with a few anchoring thoughts.
- Very few investors actually own just stocks. A balanced portfolio – say 60% U.S. equities and 40% core bonds – would have been down something close to -7.5% this week.
- From a wealth perspective, this puts diversified investors back to early Q4 2019 levels.
- Year-over-year, a balanced investor is up approximately +8.5%, with U.S. equities up over 5% and core bonds up an astounding 11%+.
- A globally diversified portfolio (proxied by the iShares Core Moderate Allocation ETF) was down just -3.9% on the week and is now only down -2.2% on the year. International diversification actually worked this time around and rates provided a decent buoy.
Source: Sharadar. Calculations by Newfound Research. Factor definitions available upon request.
Buy the dip?
I want to begin by saying that if you believe markets are a screaming buy today – more than they were last week, or even more than they were last October – you are inherently market timing. Which is fine, but should at least be recognized.
If we are going to have that view, however, we should probably at least try to reconcile where it is coming from. So let's start with the basic assumption that a stock should have equal value to the sum of all future discounted cash flows. In the Gordon Growth Model, this is written as Price = D1 / (r - g), where D1 is next year's dividend, r is cost of capital and g is the growth rate of dividends. With some simple re-arranging, we can say that r = D1/Price + g; or "return equals yield + growth."
This is often expanded to say "return equals yield + growth + re-valuation" in recognition that we often see multiple expansion and contraction over market cycles. Note that while the Gordon Growth Model specifically uses dividends, we can think of yield here as a more comprehensive shareholder yield.
What this ultimately boils down to is that for us to believe markets are a buy today, we have hold at least one of these three beliefs:
The last case can be argued due to shifts in aggregate risk aversion: as investors become more or less risk averse, valuations should go up or down.
- We believe yields will be higher than the market is pricing them;
- We believe growth will be higher than the market is pricing it;
- We believe valuations will revert positively.
The most trivial argument I have heard this week is that Coronavirus will create both supply and demand shocks in the global economy, leading to reduced growth. This may indeed be the case, particularly in the short term, but how bleak and permanent an impact must we be expecting Coronavirus to have for markets to suddenly fall over 13% in a week?
Well, if we assume yield, rate of return, and valuations stay constant, we'd have to expect a permanent shock of about -0.4% to growth rates to justify a sudden 13% decline in price.
Of course, not all of the decline has to come from an assumed shock in growth rate. It may simply be that the aggregate appetite for risk has declined, leading to lower valuations. This change in risk appetite could be temporary or it could be permanent.
But to say "today is a day to buy" (or sell!) – especially if we weren't saying it last October – requires a view as to where the market got it wrong.
The Takeaway: A call to "buy the dip" is inherently a market timing call and should be recognized as such. If investors are going to make such a call, we believe it is important for them to consider where they believe the market is mis-pricing future expectations: yield, growth, or risk appetite.
So who was actually selling (and who was buying)?
This is always an interesting question. Who is actually doing all this selling?
Markets are complex and dynamic environments and I firmly believe there is never just one answer. It is entirely possible that Coronavirus was a catalyst, but we should acknowledge that the market had reached very extended short-term levels. For example, the S&P 500 was over 12% higher than its 200-day moving average. The prior time it reached such an extended distance was late January 2018, when it hit 14.5%. So it might have been a combination of a reduction in risk appetite met with little marginal buying demand.
And then once markets start moving, all sorts of dynamics come into play. Beyond a rise in risk aversion, we might see influences from systematic de-risking, hedging in structured products, and tax loss harvesting, all of which can create a self-reinforcing imbalance between buyers and sellers.
The role of dealers and banks is particularly interesting to me in times like these. When investors buy put options (either explicitly or in the form of a structured products like buffered notes), there is someone on the other side of that trade. Options dealers may be willing to sell the put, but may not actually want to make the directional bet the put implies.
Thus, they hedge their position. In the case of being short a put option on the S&P 500, a very simple hedge would be to short some S&P 500 exposure.
As the market sells off, the puts become more valuable, and in an attempt to remain neutral to the trade, dealers must sell short more equities or futures. In option parlance, the important variable here is gamma, which measures how the option's price accelerates due to changes in the underlying security's price.
The acceleration is an important aspect here. As prices fall, the acceleration means I have to keep shorting an increasing amount to maintain my hedge. I think it's clear why this becomes a self-reinforcing trade.
SqueezeMetrics seeks to track this aggregate gamma exposure with a proprietary measure called GEX. GEX reflects the number of dollars that have to come to market to hedge a +/- 1% move in the S&P 500. A positive GEX implies that a broad positive gamma exposure in the market, implying that hedging orders would come to market in the opposite direction price moves. i.e. As price goes up, hedgers sell and vice versa.
We can see that GEX turned negative this week, implying that hedging orders would come to market in the same direction of price movement. On Thursday, this value hit -$2.96 billion, implying that for every 1% the market fell, another $2.96 billion would be sold short by hedgers.
On the other hand, we should consider that banks might be taking the exact opposite trade due to their structured product positioning. According to StructuredRetailProducts.com, a significant proportion of institutional structured product flow is into reverse convertible notes.
In this structure, the client has effectively sold a put option to the bank. The bank, therefore, is implicitly short the underlying index. To hedge this exposure, the bank will buy the underlying exposure. However, these notes are most often sold on price returns, whereas the bank is hedging with a total return exposure. The difference: dividends.
To get rid of this dividend overhang, banks turn to the dividend futures market and sell their exposure short. We may see evidence of this in the extreme flattening of the futures curve over the last week.
(Thanks to Eric Ervin for highlighting this for me. If you want to learn more about this market, I recommend listening to my conversation with Eric Ervin, the manager of the only ETF that I'm aware of that gives access to it: DIVY. I would also like to thank Benn Eifert, who entertained at least several hours worth of questions about the dynamics of this market during my flight home from Chicago on Friday night and Saturday morning.)
Of course, the dynamics here – like most markets – are very complicated, as long-dated dividend futures also exhibit significant, and asymmetric, equity beta (naively, one might say that an infinite length dividend swap is just equity). So it should come as no surprise that the 10-year futures have substantial equity beta and therefore as equity markets fall, this futures contract falls.
That said, the long-dated nature of structured products means they have relatively small gamma, so buying pressure may have been swamped by short-term options exposure. But given the tiny nature of the the dividend futures market, small shifts in marginal demand can lead to large shifts in price. Just more evidence, in my opinion, that hedgers were very active in all markets.
The Takeaway: Markets are really, really complicated with a tremendous amount of participants who may be buying and selling for reasons that have absolutely nothing to do with prevailing headlines. Markets going down day after day may have absolutely nothing to do with risk aversion and may merely be a function of forced hedging.
Whichever way the trend blows... Doesn't really matter, to me.
Trend is one of the primary styles we employ at Newfound. It likely goes without saying, but trend signals took a sharp turn south this week for global equity markets. Interestingly, however, intermediate- and long-term total returns for the S&P 500 remain positive.
In fact, our interpretation of whether the trend in U.S. equities is currently positive is largely dependent upon the specification of the model and parameterization. For example, models defined by whether price is above or below a moving average – which mathematically tend to overweight more recent returns versus a standard total return model – have largely shifted off. Conversely, trend models defined by a dual moving average cross-over (e.g 50 / 200 cross) have largely stayed positive, as they tend to overweight prior returns versus more recent returns.
As a more concrete example, consider that price has fallen below the 200-day moving average and the 10-month moving average (popularized by Meb Faber), but the prior 12-month return remains positive. Given the number of popular "do it yourself" tactical allocation models that rebalance at the end of month, we suspect lots of investors have some tough decisions to make this weekend. This is, potentially, the perfect storm of specification risk and timing luck...
Our position on the situation should come as no surprise: an ensemble of signals implemented in a transitional manner (so as to avoid rebalance timing luck) would likely advocate that investors begin systematically reducing equity beta. A blend of signals from the table above would suggest a target equity allocation of about 60%. Global equity trend signals display similar characteristics.
Evaluating U.S. equity sectors with a cross-section of different trend models and speeds, we can see that trends have turned sharply downward.
While specific portfolio transitions will be mandate dependent, these current trend readings would suggest a transition towards holding 50%+ in short-term U.S. Treasuries in our long/flat portfolios. Given how close many of these signals are to their trigger points (especially in the price-minus-moving-average models), a quick rebound could put many of these signals back into positive territory.
A particular curiosity in the equity space this week was the strange behavior in traditionally defensive sectors and factors towards week's end. Relative performance of Utilities, Consumer Staples, REITs, and Minimum Volatility all dramatically under-performed the S&P 500 on Friday. Interestingly, the sell-off on Friday was so extreme in these exposures that all but Consumer Staples ended up under-performing broad equities for the week.
Source: Tiingo. Calculations by Newfound Research.
Given how precipitously U.S. Treasury rates fell this week – including Friday – this result is somewhat of a head-scratcher. I'm all ears if anyone has a theory on this one.
The Takeaway: Trend signals are largely mixed depending upon the speed and model under investigation. An ensemble approach, based upon current signals, would suggest transitioning over time to a partially hedged (approximately 50%) allocation.
Where are all the TNX 1.20% hats?
Speaking of Treasury rates, the 10-year U.S. Treasury rate reached a new all-time low this week, falling from 1.471% to 1.124%. As a riff on the "Dow XX,000" hats, I made my own "TNX 1.30%" hat for an event I was attending in Chicago. It was irrelevant within 24 hours.
Equally interesting was that the yield curve actually steepened; i.e. the front end of the curve fell more dramatically than the back end, with 2s/10s steepening by 12 basis points.
This may reflect a growing expectation of imminent action from the Federal reserve. According to calculations from JP Morgan, Federal Fund Rates futures are now pricing in a total of 95 basis points of easing in 2020. In the same note, JPM published the following table on speculative positioning in interest rate futures. Curiously, this data suggests a bifurcation, with an extension of longs at the front-end of the curve and increasing shorts at the back end.
Based upon this positioning, short covering could drive longer yields down significantly.
Within our own portable beta U.S. Treasury futures strategies – which we apply in our tactical equity mutual funds – we evaluate three primary signals: momentum, valuation, and carry.
It should come as no surprise that the term structure of momentum remains positive for short-, intermediate- and long-term measures, suggesting that trend followers will continue to be net buyers of these positions.
Source: Stevens Analytics. Calculations by Newfound Research.
Valuation tells another story, however. Here, we prefer to measure deviations of real yield from a long-term average. And if we though bonds were expensive last August ... well, they haven't gotten cheaper. While a z-score reading below -2 suggests a heightened risk of positive mean reversion in rates, value tends to have a long forecasting horizon. In other words, the reversion may take months, quarters, or even years to play out.
Source: Federal Reserve of Philadelphia; Federal Reserve of St. Louis. Calculations by Newfound Research.
The final signal we evaluate is carry, which is our estimate of expected excess returns, incorporating both spread and roll yield. Interestingly, while the 10-year rate declined this week, the steepening of the yield curve was actually positive for carry, increasing both the spread and the expected return from roll.
Carry is still historically quite thin, but at least remains positive.
Source: Federal Reserve of Philadelphia; Federal Reserve of St. Louis. Calculations by Newfound Research.
While we still believe the absolute level of rates represent a significant financial planning problem for investors going forward (recall that starting yields are a fantastic predictor of future returns), our signals continue to suggest mixed positioning with respect to rates in the short term.
On the credit side of the equation, option-adjusted spreads for high yield bonds climbed from 366 basis points to 462 basis points (as of Thursday). This remains well below the January 2019 peak of 544 basis points. The option-adjusted spread on US corporates only moved from 105 to 121 basis points (again, as of Thursday), significantly below the January 2019 peak of 163. That said, the speed at which spreads changed week over week was quite extreme, with change in the ICE BofAML US Corporate Master OAS hitting north of the 98th percentile.
Nevertheless, historically high durations in both IG and HY corporates helped meaningfully offset spread expansion.
On the alternative income side, it should come as no surprise that more equity-sensitive exposures – dividend equities, REITs, MLPs, and covered call strategies, and convertible bonds – took the largest hits this week. As with traditional equities, many of these positions are exhibiting mixed trend signals.
Credit-sensitive assets – including senior loans, emerging market debt, and high yield bonds – were next in line. But with only a minor hiccup in credit spreads, they were not out-sized losers this week, and thus predominately continue to exhibit positive trends.
The Takeaway: U.S. Treasury signals remain mixed: trends remain positive but valuation appears at extremes and carry is very thin. Positioning based upon a combination of these signals suggests partial exposure. A steepening curve and Federal Funds Rates futures positioning suggests the Fed will cut substantially this year. In alternative income asset classes, trends among non-equity exposures remain largely positive and we continue to see ample opportunity for diversification, even if riskier exposures are trimmed due to negative trends.
For equity-centric investors, this was a rough week. The day-after-day selling felt relentless and it can be hard to fight the urge to panic.
But taking a step back, we find that well-diversified portfolios fared much better and that, while sudden and jarring, we're effectively back to October 2019 levels.
From a quantitative perspective, most of our signals are somewhere in the "yellow light" region. Short- and intermediate-term trends have largely turned negative, but are close enough to trigger points that they could rapidly go the other way. Long-term trends are still mostly positive, with the exception of price-minus-moving-average models.
And despite strong over-valuation signals, trends in U.S. Treasuries remain positive.
These yellow lights suggest a partially de-risked position in most of our mandates, which is how we expect to transition should such signals remain constant.
While volatility typically begets volatility – and we certainly expect more in the short term – we believe this week serves as an important reminder to investors that the unexpected is almost guaranteed to happen eventually. We believe that having a plan before risk appears, and then sticking to it, is the best way to navigate these periods.
Thank you for taking the time to read this note. If you have any questions, please do not hesitate to reach out.
✉️ Join Newsletter | ✎ Send us your thoughts | ✄ Share URL