Hindenburg+ Indicator: Time To Bail?
As we close out the first half of 2017, equity markets have managed to defy gravity, mediocre economic conditions, company fundamentals and rich valuations. U.S. equity markets has prolonged a historically long period of suppressed volatility (VIX), as the S&P 500 has truly been bullet-proof. The Dow Jones Industrial Average, the Russell 2000 small cap index, and the S&P 400 mid cap index have all remained solid year-to-date.
European indexes along with the Nasdaq-100, however, have shown some signs of fragility in final month of the first half. The charts of these higher beta markets are worth noting, as they may represent the first cracks in the dike. The top chart shows the EuroStoxx 50, which has entered a downward consolidation channel. While this could turn out to be a bullish pennant figure, which will resolve itself in the direction of the primary trend (up), the probability of a downward acceleration would seem relatively high, especially as we approach the negative seasonal period of August/September.
The next chart shows the market-leading Nasdaq-100. The index has broken below the 50-day moving average, which has served as support since December. The possibility of a failure swing is setting up here. We would imagine that a close below 5600 would switch computer algo buy-the-dip programmes into sell programmes.
In this current bubblish market context, we continue to ponder the recent initial Hindenburg Omen sighting on May 4, 2017 (with confirming sighting on May 31 and June 20). The Omen was triggered on both the NYSE and Nasdaq exchanges. Recall that Hindenburg Omen’s basic function is to identify when significant bifurcation is taking place within any given market. That is, market internals are pulling in opposite directions, leaving an absence of coherency. The original formula for calculating the Omen is:
- The number of issues in a specific exchange hitting 52-week Highs and Lows must both exceed 2.2% of the number of issues traded for the day (some use a 2.8% threshold; others, such as Bloomberg in their chart below, use 2.2%).
- New Highs are less than or equal to twice the New Lows.
- The benchmark index for the exchange must be above the value it had 50 trading days ago (ie: above the 10-week moving average).
- Once the three aforementioned events have occurred, the signal is valid for 30 trading days. During the 30 days, the signal is activated whenever the McClellan Oscillator is negative.
When the above conditions are met, a yellow point is drawn on the chart below. This is not yet the Hindenburg Omen. Two such signals within a 36-day period is considered a Hindenburg Omen and is indicated by a red diamond on the chart below. The Hindenburg Omen portends a serious decline beginning within the next 40 days.
Critics remind us that the track record of the Hindenburg Omen is spotty. Of the 42 completed Hindenburg Omens since 1985, only 9 resulted in market crashes…a 22% “success rate”. We circled the completed signals (yellow + red dots) in the chart above. We must insist on one point: a Hindenburg Omen sighting has been a sine qua non condition for each market crash since 1985.
Our analysts re-examined the criteria to try to improve on the reliability of the original Hindenburg Omen. We introduced new criteria to come up with our Hindenburg+ indicator, which vastly improves upon the success rate of the original Hindenburg. The downside of our work is that the new market data introduced into the equation limits the historical track record to 1996 due to the inexistence of data prior to this date.
Our Hindenburg+ indicator is composed of the following inputs:
- The number of issues in a specific exchange hitting 52-week Highs and Lows must both exceed 2.2%
of the number of issues traded for the day (as above).
- New Highs are less than or equal to twice the New Lows (as above).
- The benchmark index for the exchange must be above the value it had 50 trading days ago (as above).
- A negative McClellan Oscillator (as above).
- The rolling 30-day correlation coefficient between the VIX and S&P 500 is becoming “less negative”
- NYSE Real Margin Debt as a percent of Real U.S. Household Income greater than 3% (new).
- Total U.S. Market Short Interest as a percent of Total US Market Cap is lower than the level that
prevailed 263 trading days ago, or one calendar year (new).
The logic behind the S&P 500/VIX correlation is that in normal market conditions, when the S&P 500 rises, the VIX falls. When this traditional relationship begins to break-down, we estimate that something is astray. For the real margin debt as a percent of household income, when investors get too leveraged up, bad things often result. The crowd is always wrong at market extremes. Finally, the relative absence of short sellers is again a sign that everyone is crowding into the same side of the Zeppelin, so to speak.
The chart below presents the confirmed Hindenburg+ signals.
The following table analyses the quality of the above Hindenburg+ signals.
With the exception of 2006, a Hindenburg+ signal has merited selling for an active investor.
So will we see a market crash begin in the coming weeks? The answer remains – “this is anyone’s guess”. First, no one has ever had the ability to predict the magnitude of a market decline. Corrections following the Hindenburg+ ranged from -9% (mild pull-back) to -54% (severe crash). Second, unlike with the past, we have never had central banks so involved in financial markets as today. It is possible that unlimited central bank liquidity will check any severe future selling. Finally, the Hindenburg+ can only be tested from 1996, so predictions (although accurate over the past 20 years) are not robust.
In advance of the seasonally difficult period of August/September, the weight of the evidence would suggest that anticipatory profit-taking would behoove active investors. However given the tenacious market uptrend and the continued support of central banks, we see no evidence at this time for more than a garden-variety equity correction to relieve excesses that have build up since the U.S. election.
Like to give your opinion? Go to Comments