EZV Algorithms: I Smell A Bear Trap; Maintain Perspective While Others Panic

Summary

• I smell a bear trap here, but I’ll explain why if I’m wrong, it’s okay anyway.
• EZV Algorithms metrics continue to indicate ‘hold’ and the ISM data that set the market on its ear is not quite what headlines would imply.
• My goal: beat a buy-and-hold strategy by avoiding major downturns. If B&H earns 10% , I want 15%+ and this could be a lesson in how to do it.
• Probabilitiesskew toward a breakout to the high side despite transient setbacks onsuperficial news stories and the markets natural fear of new heights.
• One possible concernis that earnings projections could be falling disproportionately vs. the slowinggrowth rate. That would be a materialshift and could produce a risk-off sentiment.

The Dow is down about 800 points from Tuesday through midday Wednesday. Panic is in the air.

Perspective

The greater the panic, the more important it is to keep perspective and remember the game plan that works in volatile markets.

My game plan is this: My objective is to beat a buy-and-hold strategy ("B&H"). B&H earns about 9% price returns historically, and I know that if I can avoid the worst downturns, I can earn more than 15%. My methodology does not avoid every 3% downturn, but it does provide warnings of major downturns. All warnings are not prologue to major corrections. For every eight warnings, one of them avoids significant losses, and seven produce zero-sum noise compared to B&H. But the one in eight valid warnings, when heeded, produces a near doubling of returns with major drawdown protection.

So, the perspective that gives me comfort is this: Any news-driven gyrations are just noise and the losses (or gains) are consistent with B&H outcomes. If psychology turns toward a significant risk-off sentiment, I'll see it in the metrics and sell before it is truly damaging. That will provide the opportunity to reenter at lower prices and reap bigger rewards. So why worry!

If this market drop is the prelude to a major correction, I'll know it soon and the first few percentage points drop will look meaningless when viewed later. In fact, the way to beat a B&H strategy is to avoid major corrections, so if this is to be one, bring it on.

I find it helpful to view market drivers in three categories.

1. Fundamentals will drive markets in the intermediate and long term, but they're close to useless for market‑timing purposes.
2. Headlines and hyperbolic news coverage drive day-to-day gyrations, and they amount to noise.
3. Market psychology (risk-on/risk-off) drives multi-week cycles suitable for trading. Psychology can be affected by meaningful shifts in fundamentals, so the fundamental context has value, but the larger value comes from identifying the risk-off / risk-on attitudes as they emerge.

Working under that construct, this market action looks a lot like a bear trap to me. Fundamentals have weakened slightly, but there is no recession in sight. News has been hyperbolic, and that is the major driver of the current volatility. The real concern is stealthy - that earnings projections might be falling disproportionately vs. the slowing growth rate. That would be a material fundamental shift and could produce a risk-off sentiment that is more intense than run-of-the-mill headline risk. Finally, market psychology, as measured by EZV Algorithms metrics, is softening but holding up so far. So, I hold for now.

Whichever way this breaks, I'm monitoring market psychology via EZV Algorithms metrics and I'm ready for what might come. I still think that probabilities skew toward a breakout scenario despite transient setbacks on superficial headlines and the markets natural fear of new heights.

For a better understanding of EZV Algorithms algorithm and why I hold the perspective I do, please read "The Quant Corner" at the end of this article.

ISM Data

Much like the inverted yield curve hysteria, the ISM data is being touted as a recession indicator, so I looked into what lay beneath the headlines. The main items that troubled markets were PMI (­‑1.3%), Inventories (-3.0%), new export orders (-2.3%), and production (-2.2%). Look at them one or two at a time.

• PMI (Purchasing Managers' Index) while down 1.3%, sat at a 47.8 index. ISM reports that a 47.8 PMI correlates to a 1.8% growth rate for GDP, not a recession.
• Production was down 2.2%, but inventories were down 3.0%. I know these are survey index numbers and not true measurements of underlying quantities, but it seems to me that a bigger drop in inventories would imply higher final demand.
• New export orders were down 2.3% and I suspect that is a true reflection of issues surrounding China.

So, the ISM data is consistent with a slower growth rate, inventory changes, and trade skirmishes. Not much news there, and certainly not a near-term recession harbinger.

There is one component in the ISM data that might be more troublesome, and oddly, it registers as a positive in ISM metrics. The 'Price' metric rose 3.7% because prices are falling. ISM said this: "Prices contracted in September, but at a slower rate compared to August. Respondents reported decreases in prices for aluminum, corrugate, oil, plastics, wood pulp and steel products."

While that is another indication of deflationary pressure, it will also probably strengthen the Fed's case for further cuts.

Market Psychology

At midday Wednesday October 2, 2019, with the Dow down about 500 points on the day so far, EZV Algorithms metrics were indicating "Hold." They reflect some damage from the two-day price drop, but it still looks like the Confirming Slope will sustain an indication to hold equities. Here is the dashboard.

EZV Algorithms Dashboard

Source: Michael Gettings Data Sources: Fidelity, VIXCentral.com, CBOE

For those who have not been following my series of articles, EZV Algorithms algorithm tracks changes in the shape of the VIX futures curve. To be clear, it does not simply track the VIX, and it does not simply measure the contango or backwardated shape of the VIX term structure. It focuses on the rate of change of the VIX term structure to identify when emerging equity risk merits a sell signal. The SHAPE represents the average contract-to-contract discount for the first four futures contract months (3 deltas).

More explicitly, when the VIX term structure's rate of change (Slope), measured over two different look-back periods, is moving toward backwardation rapidly enough it signals "sell."

For new readers, at this time the Confirming Slope graphic is the critical item in the dashboard because the slope is materially positive and, given the pattern of data, it is likely to remain so for a while. Individual data points can change quickly, but it will take a few very negative data points to drive the Confirming Slope below its trigger level.

To wrap up this discussion I'll reiterate my view on all this, I'm not worried because news-driven gyrations are just noise and the losses (or gains) are consistent with B&H outcomes; if a big risk-off sentiment comes I'll see it and exit to buy back in at lower prices. Meanwhile I hold while monitoring metrics.

Housekeeping

I'm working with Seeking Alpha to set up a Marketplace service, so if you're interested in getting access to daily metrics, more frequent posts and numerous other features, keep an eye out for EZV Algorithms launch date and discounted pricing for early subscribers. And thanks for following.

The Quant Corner

EZV Algorithms algorithm measures the rate of change in the shape of the VIX term structure. The SHAPE metric is the basic building block. A contango shape is represented as positive values (near-term VIX futures contracts trade at discount to longer-dated contracts). Generally, a contango shape is consistent with a risk-on attitude, but the SHAPE metric alone does not tell the story. The opposite of contango is backwardation (near-term contracts trade at a premium to longer-dated contracts), and that condition is consistent with a risk-off attitude. But once again, the SHAPE holds little predictive value by itself.

The algorithm measures the rate of change in SHAPE over two different look-back horizons. A short look back (Primary Slope) provides transient sentiment changes, while a somewhat longer look back (Confirming Slope) provides a more stable trend. When both Slope metrics fall below trigger levels, the algorithm indicates action - Sell the equity ETFs and buy Treasuries (IEF) - with one exception. If the SHAPE is above a threshold "safe" level, the sell signal will be negated. When those Slopes move back above trigger levels, a reentry signal is posted.

Here is a key to interpret the algorithm:

Importantly, look-back periods and trigger criteria are set by an artificial intelligence methodology that derives those parameters, applying them to prospective periods only and never to any periods from which they were derived. In other words, over fitting should not be an issue.

Performance comparisons reflect holding versus trading an ETF portfolio of SPY, DIA, QQQ, IWM and SSO with QQQ over-weighted, IWM under-weighted, and SSO representing 10% of the portfolio. Trading signals come from the algorithm.

Modeling since 2008, produces substantially improved drawdown performance and superior returns. The drawdown protection is the primary goal; the algorithm is a risk-mitigation tool. Yet avoiding big downturns produces markedly better returns by compounding on a higher investment base. Here is a table of drawdown performance for various terms over eleven-plus years of study.

Drawdown Protection for Varying Terms (252 trading days equals one year)

Term	B&H Worst G/L	Algorithm Worst G/L	Worst Drawdown Advantage
252 days	(37.0%)	1.1%	38.2%
126 days	(47.9%)	(13.7%)	34.2%
63 days	(45.0%)	(23.7%)	21.3%

Click to enlarge

Source: Michael Gettings Data Sources: Fidelity, VIXCentral.com, CBOE

Avoiding big drawdowns produces substantially better returns thanks to the power of compounding. Here is a graphic of performance since April 2008. B&H returns were 9.4% and EZV Algorithms returns were 17.4% but the effects of compounding are dramatic.

EZV Algorithms Performance versus Benchmark

Source: Michael Gettings Data Sources: Fidelity, VIXCentral.com, CBOE

One more technical note: Originally the algorithm was a bit late in calling reentry opportunities, so the Primary Slope parameter is now triggered, for reentry only, on a next-day forecast value. The recent accuracy of those forecasts is shown below.

Next-Day Primary Slope Forecast Accuracy

Source: Michael Gettings Data Sources: VIXCentral.com, CBOE

The forecast methodology produces reasonably accurate one-day forward estimates, particularly for such a volatile factor; while not perfect, the accuracy is adequate to produce a noticeable improvement in the eleven-year performance.

Thanks again for reading and especially for following.

Disclosure: I am/we are long SPY. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.