Drops, Pops, Streaks and Flips

Bottom Line: Two down days and one up in the up issues ratio has served as an effective “buy” signal over the past 40 years.

There are a few catchy phrases in market lore about Federal Reserve interest rate changes such as “three steps and a stumble” and others of the sort. I’m going to show something today related to breadth that we’ll call “two drops and a pop”, and it has been a perfect buy signal for the past 10 years. In fact, it has been nearly perfect for the past 40 years, with only two or three (depending on your time frame) losing trades out of fourteen occurrences. Out of all the trades, really only one was what I would consider a “bad” signal.

The breadth figure I’m talking about is the up issues ratio. This is simply the number of advancing stocks divided by the total number of advancers and decliners. For example, on Friday I show that on the NYSE there were 2402 stocks that rose and 887 that declined from the prior day on a close-to-close basis. So Friday’s up issues ratio was 73% (2402 / (2402 + 887)). By “two drops and a pop”, what we require is that the up issues ratio be 30% or less for two consecutive days, then the next day is 70% or greater. This gives us a sequence of two breadth thrusts to the downside, then one to the upside. And over the past 40 years, such a sequence has been a beautiful buying opportunity. Before anyone gets too excited, however, I want to caution that going back even further – to the past 65 years – the signal lost some of its effectiveness. It was still very good, but the first 25 years were not as good as the last 40 years.

The table below outlines the performance in the Dow Jones Industrial Average the given number of days after each occurrence of the “two drops and a pop” for the past 40 years.

“Two Drops and a Pop” 1963 - 2004
Date	5 Days	10 Days	30 Days	60 Days	90 Days	120 Days
11/26/63	1.6%	1.8%	4.1%	7.2%	10.7%	9.9%
05/18/66	1.4%	0.5%	-1.0%	-4.3%	-9.8%	-8.7%
03/06/68	0.6%	-0.8%	7.2%	8.1%	5.8%	10.1%
05/27/70	11.1%	8.0%	7.8%	13.5%	19.2%	18.2%
07/08/70	4.3%	6.2%	6.1%	11.5%	12.6%	21.8%
10/13/76	0.7%	0.8%	0.9%	4.1%	-0.9%	-3.4%
01/04/80	3.6%	4.6%	6.8%	-5.2%	-1.4%	7.1%
09/30/80	3.0%	3.2%	3.5%	3.6%	1.6%	6.8%
11/17/82	-2.7%	0.5%	1.9%	5.7%	10.3%	19.7%
10/21/87	-8.9%	-4.1%	-12.4%	-3.2%	2.1%	3.9%
04/09/92	4.4%	3.1%	5.0%	2.2%	3.2%	1.4%
07/17/96	-0.4%	2.8%	6.2%	10.1%	19.4%	22.8%
09/24/01	2.7%	5.4%	9.7%	16.2%	15.1%	22.9%
08/06/02	2.5%	7.2%	-1.2%	1.9%	3.2%	-2.2%
Avg. Return	1.7%	2.8%	3.2%	5.1%	6.5%	9.3%
% Positive	79%	86%	79%	79%	79%	79%

From the table, we can see that after 10 days, only two instances showed negative returns – March 1968 and October 1987. However, both of those ultimately proved to be excellent bottoms as the Dow rose handily over the next year. The one “bad” signal out of all of these was in May 1966, at least after an initial rally. This one occurred after one of the only real corrections after the bull market of the early 1960’s. When the signal happened in May of ’66, there was a bounce over the next few weeks which proved to be a major head-fake as the Dow dropped off a cliff for the next three months after that. Looking at all the data back to 1940, there was only one other instance that I would consider to be bad, and that was in June 1946. Once again, this happened after one of the first real corrections after a multi-year bull market, and the head-fake rally in the few weeks after the “two drops and a pop” buy signal lead to a waterfall decline into September of that year.

The table below shows the average performance in the Dow the given number of days after every occurrence of the signal since 1940.

Whenever we talk about breadth figures anymore, we have to at least mention the fact that decimalization has forever changed our comparisons to historical periods when fractions were still used. However, I think using ratios like we did here minimizes the impact of decimalization, since we are comparing issues that rose on the day to all changed issues. So while there may be more “up” issues, there also are more “down” issues, and taking a ratio of one in relation to both (instead of one all on its own) alleviates some of the distortion. Even if you want to discount the excellent results from the last 40 years, the fact that this signal has done so well over the past 51 signals since 1940 should not be ignored. We’re always using probabilities here, and as we saw from a couple of the instances, there is absolutely no guarantee that the market will rise for the next six months simply because that’s what the average has done. But when you look at it from a risk/reward perspective, the odds according to “two drops and a pop” are skewed to the upside.

Bottom Line: When the VIX shows back-to-back extreme closes, it has lead to short-term weakness, longer-term gains.

In an intraday comment Friday, I mentioned a statistic about large one-day declines in the CBOE Volatility Index. The stat was that when the “old” VIX (current ticker: VXO) closed 10% or more below where it opened that day, then the S&P 500 futures made a higher high the next day 78% of the time (by 0.6% on average). In addition, the next day's low in the S&P was higher than today's low 88% of the time (meaning there was only a 12% chance of the next day dipping below today's low). We narrowly missed the 10% mark on Friday, as the VXO declined just over 9% from where it opened.

One other aspect that I want to share is about those times when we saw large back-to-back volatility swings, meaning a 10% or greater rise in the VXO one day, then a 10% or greater drop the next day (on a close-to-close basis) - this is what happened on Thursday and Friday of last week. Since inception in 1986, this has happened on only 27 occasions not including last week. The table below shows how the S&P 500 futures reacted after those instances:

We can see from the table that after previous bouts of extreme volatility swings, there was generally some slight weakness immediately after. However, that soon subsided and the market reacted quite positively afterwards. After 60 days, the futures were higher 85% of the time, well above a random period. If we add in a crude trend filter, such as whether or not the S&P 500 futures were trading above or below their 200-day moving average, then the pattern from above remains. There were 15 such occurrences, and the futures closed higher the next day only 6 times, but after three days the futures were higher 11 out of the 15 times. After 30 days, the futures were higher 13 times and after 60 days they were higher every time but once.

Bottom Line: Historically, when the TRIN has closed above 2.0 for 3 out of 5 days, it has lead to some very positive performance.

One of the most popular subjects over the past couple of days has been the persistently high TRIN readings seen on the NYSE. For three days in a row, we saw the TRIN close above 2.0. A streak of three straight days above 2.0 is extraordinarily rare – in fact it has only happened twice in the past 64 years. The first instance was the three days ending 8/2/43 after which the Dow Jones Industrial Average went on to immediately rally 5% over the next month and a half before rolling over and bottoming for good in November. The other instance was the three days ending 9/14/53 which marked the exact low, after which the Dow went on to a tremendous run of doubling over the next three years. Impressive precedents, to say the least.

Two precedents makes it difficult to conclude anything concrete, so we need to relax the parameters a bit to see if any more time periods qualify as being comparable to what we saw last week. Instead of looking for streaks of three days in a row with a TRIN over 2.0, let’s instead look for any period where the TRIN closed that high for 3 out of 5 days. Using those criteria, we come up with 12 other periods since 1940. The table below shows how the Dow performed the given number of days after such instances.

Pretty positive results here, especially after 60 days, when 9 out of the 12 occurrences put in positive returns. If we look at those times when the Dow was in a positive market at the time, broadly defined as being above its 200-day moving average, then we only get three occurrences – and the last one happened 53 years ago. After those three instances, the Dow was higher 60 days later every time, with an average gain of 4.7%.

The biggest question that must be addressed is whether decimalization is going to make it easier for us to see these types of streaks than it would have in the past, when it was more difficult to move a stock out of the “unchanged” category. Since decimalization was introduced in 2001, there have been 75 days showing a closing TRIN reading of 2.0 or above. This is the largest number of any equivalent number of days since 1940, with the only comparable period being 1946 – 1949 when we saw 57 days with a TRIN over 2.0. I’ve looked at this a number of different ways, and it is clear to me that it is easier now than it has been in the past to see high TRIN readings. This in itself must make any comparison to historical periods somewhat suspect. While I believe the readings from last week can still be considered a positive for the market, it results should be “asterisked” due to the decimalization issue.

At the end of this coming week, March contracts for futures and options expire. The next two weeks exhibit a quite consistent seasonal pattern, in that the week of expiration tends to be positive, while the week after tends to be negative. Over the past 103 option expirations, beginning in 1995, the S&P 500 has gained 172 points the week of expiration (with 66% of them being positive), while the week after expiration the S&P has lost 185 points (with 47% of them being positive).

If we are in the process of forming a multi-week low, it would be relatively unusual to see it come in the form of a straight move down and straight move up again. Often, the low made on a short-term panic move is tested a few days later. I would qualify the action we saw last week as panicky, as determined by many of the measures that we follow (on a short-term basis only). This seems almost too scripted, but the perfect pattern to see would be a generally rising market next week, in keeping with the usual option expiration pattern, then another decline post-expiration that tests the low put in late last week accompanied by positive divergences in the technical indicators that so many traders follow. Should that happen and the market begin to recover, I would look to become a very aggressive buyer.

As it stands, however, I am cautious. The extreme oversold readings we saw last week look to me like the overbought readings we saw last March, only in reverse. During longer-term trend changes, we very often see short-term extremes in the direction of the trend change. That is why I have been stating for months now that it is imperative to watch how the market rallies from short-term oversold conditions – if it cannot rally, then we have a decent clue that the larger trend is changing. If the uptrend had held true to form, by rights we should have rallied Wednesday and definitely Thursday. I can see a little leeway due to the terrorist attacks in Spain, but downside should be limited IF the uptrend is to continue. If we decline below Thursday’s low and hold there, then I would rather be short than long.

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.