Omen, or "Oh, Man!"?

http://www.sentimentrader.com

Omen, or “Oh, Man!”?

Tuesday, April 27th, 2004 10:45pm EST

An Undeserved Omen

Bottom Line: The “Hindenburg Omen” is one of those pieces of market lore, predicting a crash, that more appropriately belongs on the bench.

Something I’ve been asked about a few times in the past week is an ominous-sounding signal called the “Hindenburg Omen”. Getting background information on its discovery is difficult at best, and quite frankly I’m not sure to whom it should be attributed. I’ve seen references to Norman Fosback, Ian McAvity, Kennedy Gammage and – I’m serious – a blind math genius in Florida by the name of Jim Nieska. Just as there are varying interpretations of who invented the signal, there are variations on how it should be computed. To simplify things, I’ll just go with Ian McAvity, whose suggestion is to look at a 5-day moving average of new highs and new lows on the NYSE. If both moving averages are greater than 2.4% of the total number of stocks traded, then we have a Hindenburg Omen. The theory is that a great deal of uncertainty in the market, characterized by a large number of new highs and new lows at the same time, occurs before major market meltdowns. As usual, this is true (but only a little bit).

The usual examples given for this phenomenon are the ones that occurred before the crash in 1987, before the sharp drop in 1990, before the mini-crash in 1998, near the top of the bubble in September 2000, before 9/11 in September 2001, and lastly before the waterfall decline in July 2002. Just looking at those examples, it makes one shiver – it’s uncanny how the market slid precipitously soon after such signals. Unfortunately, or fortunately depending on how you look at it, that’s only half the truth. Typically, examples that go against the theory are simply left out, like the one at the low in March 2001, or the low in October 1998, or the low in December 1991, or – and this is the best one – the one that occurred very near the ultimate low in July 1982. It is coming up now because the signal occurred nine days ago on 04/14/04.

Instead of giving you my opinion on what this signal may or may not mean, below is a table showing every Hindenburg Omen since 1965.

Hindenburg Omens 5-Day New Highs and 5-Day New Lows Both at Least 2.4% of Total Issues 1965 - 2004
Date of Signal	New Highs	New Lows	10 Days Later		30 Days Later	60 Days Later	90 Days Later	120 Days Later	252 Days Later
03/31/65	3.7%	2.4%	2.4%		4.8%	-3.6%	-0.3%	4.2%	3.0%
11/23/65	5.7%	2.4%	-0.5%		1.4%	1.0%	-2.0%	-8.0%	-13.2%
05/31/67	2.7%	3.0%	3.7%		3.7%	4.5%	9.2%	2.9%	12.2%
10/11/67	6.3%	2.5%	-1.9%		-2.6%	0.1%	-5.3%	-2.6%	9.1%
03/29/68	2.6%	6.8%	7.1%		8.8%	10.2%	9.4%	14.8%	16.5%
03/27/69	2.6%	6.0%	0.5%		3.7%	-3.7%	-7.0%	-5.8%	-11.3%
09/22/69	2.5%	5.0%	-2.4%		1.6%	-6.2%	-10.4%	-8.1%	-13.6%
03/25/70	2.6%	5.2%	-1.4%		-11.1%	-14.8%	-13.1%	-8.6%	11.7%
05/13/71	2.7%	2.4%	-3.2%		-4.6%	-8.9%	-3.3%	-9.3%	2.7%
10/06/71	2.7%	2.5%	-4.2%		-7.0%	2.3%	5.3%	7.5%	10.5%
03/22/72	3.2%	2.7%	2.5%		-0.6%	1.4%	0.5%	1.5%	2.8%
10/26/72	2.5%	3.3%	2.5%		7.3%	4.9%	2.6%	-0.9%	-1.5%
01/16/73	2.8%	2.8%	-1.8%		-6.0%	-4.7%	-8.6%	-12.4%	-19.0%
01/03/74	2.6%	3.0%	-2.5%		-8.9%	-5.8%	-9.2%	-10.8%	-29.6%
05/25/76	2.6%	2.4%	-0.8%		4.5%	3.9%	4.7%	0.4%	-1.8%
03/18/77	4.6%	2.8%	-2.6%		-2.9%	-2.0%	-3.0%	-5.4%	-10.8%
05/25/78	3.3%	2.5%	3.2%		-1.6%	7.3%	6.0%	-4.5%	3.2%
10/04/79	4.6%	2.5%	-6.0%		-5.5%	-2.0%	7.0%	-10.4%	16.3%
07/16/82	2.7%	2.9%	-3.6%		5.4%	21.1%	20.8%	27.8%	49.5%
07/23/86	2.6%	2.7%	-0.8%		6.4%	0.4%	4.4%	8.9%	29.2%
09/28/87	2.6%	3.1%	-4.3%		-24.8%	-22.7%	-22.0%	-16.1%	-16.8%
10/16/89	2.5%	4.4%	-2.3%		0.9%	1.7%	-5.5%	-0.8%	-11.6%
07/12/90	3.5%	2.4%	-2.6%		-16.0%	-14.8%	-13.2%	-10.7%	3.2%
12/06/91	2.5%	2.5%	2.1%		8.8%	8.0%	9.8%	9.6%	14.0%
02/23/94	2.9%	2.5%	-0.8%		-4.2%	-3.4%	-5.2%	-1.2%	3.4%
10/26/95	3.5%	2.4%	2.9%		7.1%	6.3%	13.1%	11.6%	21.8%
12/15/97	3.8%	2.6%	0.8%		2.3%	10.9%	12.8%	16.1%	20.7%
06/26/98	2.5%	3.2%	2.8%		-4.4%	-9.1%	-2.0%	2.5%	17.5%
10/05/98	2.4%	7.6%	7.5%		14.9%	24.6%	24.4%	32.5%	31.6%
04/12/99	2.5%	2.7%	0.1%		-3.8%	2.7%	-1.9%	-5.6%	11.6%
11/17/99	2.4%	3.2%	-0.1%		4.1%	-1.5%	6.9%	-2.0%	-1.5%
09/19/00	3.5%	2.5%	-2.3%		-2.1%	-6.8%	-6.6%	-18.0%	-31.0%
03/20/01	2.6%	2.5%	-3.2%		10.9%	6.8%	5.5%	-9.1%	0.2%
09/06/01	3.1%	3.0%	-9.0%		-1.9%	5.5%	1.2%	5.1%	-19.6%
06/24/02	3.0%	3.0%	-4.0%		-13.4%	-12.4%	-10.3%	-9.2%	-0.9%

		Average Return	-0.6%	-0.7%		0.0%	0.4%	-0.4%	3.1%
		Standard Deviation	3.4%	8.1%		9.5%	9.8%	11.4%	17.3%
		% Positive	37%	49%		54%	49%	40%	60%

My apologies to those who have to squint to see the numbers, but formatting restrictions required a small font for this table. The columns show, in order, the date the signal was first issued (note: since we usually see many days in succession, I have shown only the first day when the signal was generated, and any occurrences happening within two months of another one have been eliminated, to get rid of “double-counting” to some degree), the 5-day moving average of new highs as a percentage of total stocks traded, the 5-day moving average of new lows as a percentage of total stocks traded, and the performance of the S&P 500 index 10, 30, 60, 90, 120 and 252 days later.

There are a couple of things to note. After 10 days of the initial signal, the S&P was higher only 13 out of 35 times, with an average loss of 0.6%, so there may be a bit more negativity than usual there. Also, in all time frames (even out to a year) the average return is extraordinarily small, which is why I thought it would be instructive to include the standard deviations of returns, which are large. What does that tell us? It shows that while there may not be a distinct bias to S&P performance after these signals, the market does tend to move well in one direction or the other.

Perhaps there are some “tweaks” to this signal that I don’t know about (such as requiring that 10 out of 30 days shows a signal, or that the S&P has to trade below its 200-day moving average, or something similar), but as it stands I don’t see a whole lot of forecasting ability for this signal. I certainly don’t think it’s a bad “omen” for the market, as it has kicked off its fair share of excellent rallies as well as waterfall declines. It does seem to coincide quite often to the beginning of a major move within a month or so, but there is really no telling if the move will be up or down. I think it may be better to toss this one into the “sounds good in theory, but…” file.

Stealthy STEM

Bottom Line: Our short-term STEM model has crept into overbought territory, a foreboding signal over the past six months.

Our STEM model (Short-Term Extreme Model) is based primarily off of four intraday measures – the VIX, put/call ratios, TRIN, and cumulative TICK. While it can and does get “stuck” at the beginning of strong, intermediate-term trend changes, it tends to flourish during times of more moderate moves and trading ranges. Over the past six months, it has been a handy guide to the short-term extremes we’ve seen. The chart below is a current snapshot:

The red arrows on the chart highlight those times the model approached 20% or below and began to reverse, which we’re perilously close to touching as of today’s close. Out of the 7 instances, there was really only one “failure”, which was the last one occurring on March 31^st, as we were rallying from the extreme oversold conditions at the time. The greatest (and most consistent) losses in the S&P 500 came after two days, with an average decline of just over 1%. Even after 10 days, the S&P 500 was lower 6 out of 7 times, with an average loss of 1%. Certainly these weren’t precursors to large declines, but considering the overall uptrend during the period, I think these results should at least be noted. At some point, this model will fail horribly as we get stuck in the beginning thrust of a new trend, but in the meantime it is suggesting that the upside may be limited in the short-term.

Stuck in Neutral

Bottom Line: Every one of our short-term “score” indicators is in neutral territory, a rare occurrence that should trigger a decent move.

While the STEM model has crept down into overbought territory, most of our shortest-term measures are neutral. In fact, we are seeing a rare occurrence today in that every one of the 14 indicators that make up the short-term indicator “score” on the site is in neutral territory. In the past two years, this has only happened three other times, highlighted in yellow below.

After the last occurrence, 7/30/03, the S&P gained about 17 points intraday before finally losing 27 points over the next 5 days. After the previous occurrence, 1/16/03, the S&P wasn’t able to gain anything and instead lost 55 points during the next week. The only other instance, 12/23/02, saw the S&P gain just over 6 points but lose 28 points over the next 5 days.

I’m not sure there’s much we can comfortably conclude about such occurrences, since there are only three of them, but I do think it highlights an equilibrium point that should bring a trade-able move in one direction or the other very soon.

Conclusion

As if we need another display of uncertainty in the market, today there was a total of just over 82,000 options contracts on QQQ and OEX traded on the CBOE – one of the lowest non-holiday-related options volume totals in the past couple of years. In fact, it was about ½ the average volume over that time. This meshes well with our neutral short-term indicator score mentioned above, as well as our previous discussion of neutral respondents in the lowrisk and Investor’s Intelligence surveys.

The action in the broader market over the past few days has alleviated any oversold condition in our shorter-term measures, and has also worn off a good part of the oversold longer-term breadth measurements as well. The facts that we still haven’t been able to rally above resistance, and that our STEM model is seemingly so negative, are a bother to me. I have preferred to trade from the long side as long as we remain above 1120ish on the S&P 500 cash index, and generally that’s still true, but I am becoming increasingly concerned that we have not been able to make more headway. For short-term positions, being flat seems to be the best bet at the moment, until the indices tip their hand. Once they do, there are enough fence-sitters out there that we should be able to see a decent move. I continue to believe that longer-term long positions are in good shape, but again if we lose that 1120ish level, I would be backing off.

Jason Goepfert

President and CEO

Sundial Capital Research, Inc.

Disclosure: no positions

This disclosure is not intended as trading advice in any form. It is meant as a note to subscribers that the author may have a position directly affected by the market outlook reflected in the commentary. Although the author takes great pains to remain objective in any commentaries, it is only fair that readers should know that the author may have taken positions in accordance with his market outlook. Positions can and do change at any time, without notice to the reader.