Thursday, December 4, 2014

The hidden message of the Hindenburg Omen

There it is again. The "dreaded" Hindenburg Omen reared its "ugly" head again (chart via Tomi Kilgore at Marketwatch):

The Hindenburg Omen has cried wolf so many times that it has become a subject of derision. The best reaction that I have seen so far comes from Urban Carmel:

Does this latest occurrence of the Omen indicate the ending of the story where that the wolf is comes, while the villagers stay home? To answer that question, let's examine the elements of the Hindenburg Omen (via Wikipedia):
These criteria are calculated daily using Wall Street Journal figures from the New York Stock Exchange for consistency. (Other news sources and exchanges may be used as well.) Some have been recalibrated by Jim Miekka to reduce statistical noise and make the indicator a more reliable predictor of a future decline.
  1. The daily number of NYSE new 52 week highs and the daily number of new 52 week lows are both greater than or equal to 2.8 percent (this is typically about 84 stocks) of the sum of NYSE issues that advance or decline that day (typically, around 3000). An older version of the indicator used a threshold of 2.5 percent of total issues traded (approximately 80 of 3200 in today's market).
  2. The NYSE index is greater in value than it was 50 trading days ago. Originally, this was expressed as a rising 10 week moving average, but the new rule is more relevant to the daily data used to look at new highs and lows.
  3. The McClellan Oscillator is negative on the same day.
  4. The number of New 52 week highs cannot be more than twice the number of new 52 week lows (though new 52 week lows may be more than double new highs).
The traditional definition requires each condition to occur on the same day. Once the signal has occurred, it is valid for 30 days, and any additional signals given during the 30-day period should be ignored, or one signal does not mean very much, but more than one is a confirmed signal, or five or six are even more important. During the 30 days, the signal is activated whenever the McClellan Oscillator is negative, but deactivated whenever it is positive.

The hidden message of the Omen
The Hindenburg Omen indicator has a lot of moving parts and it is therefore confusing. I believe that the most important message in the Hindenburg Omen is the expansion of both new highs and low, indicating divergence among stocks and points to market indecision.

A recent post by Brett Steenbarger puts the new high and low expansion in a slightly different context (my interpretation, not his) in a way that did not refer to the Omen (emphasis added):
Truly outstanding has been the plunge in my measure of correlation among stocks, which looks across both capitalization levels and sectors. Indeed, this is the lowest correlation level I have seen since tracking the measure since 2004. Correlation tends to rise during market declines and then remains relatively high during bounces from market lows. As cycles crest, we see weak sectors peel off while stronger ones continue to fresh highs. As those divergences evolve, correlations dip. Right now we're seeing massive divergences, thanks to relative weakness among raw materials shares (XLB), energy stocks (XLE), regional banks (KRE), and small (IJR) and midcap (MDY) stocks. Why is this important? Going back to 2004, a simple median split of 20-day correlations finds that, after low correlation periods, the average next 20-day change in SPX has been -.33%. After high correlation periods, the average next 20-day change in SPX has been +1.43%.
Does this mean that the stock market is going to crash?

No. First and foremost, I would point out that the first appearance of the Omen is only a warning. It needs to be confirmed by a second appearance within 30 days to be a valid bearish signal. That hasn`t happened yet.

However, Steenbarger's analysis does show that this latest episode should be regarded as a potentially bearish market factor to keep an eye on. The divergence between the major averages and the resource sector can largely be explained away by the precipitous fall in oil prices. Whether you believe that is an exception to the correlation divergence bear case is up to you.


Anonymous said...

I was first introduced to HO from the albertarocks website.Never saw the stipulation that the McClellan had to stay negative or its wiped out.This must be strictly his interpretation on this.

I Will Never Accept The Terms of Service said...

Here's another problem with the Hindenburg Omen. It uses the NYSE as a measure. A large part of the NYSE issues are non-stocks. So NYSE breadth doesn't measure what people think it does.

CanopyTree said...

What underlying assets does the NYSE measure that are not stocks?

Anonymous said...

@canopy tree: there are a good number of closed end funds (many of them bond funds), master limited partnerships, etc listed on the NYSE.