Low Behavior

Bottom Line: There is a correlation between a rise in rates over a 10-day period and negative performance in bank stocks 60 days later.

Most traders, whether they are technical daytraders or fundamental long-term investors, watch the interaction between various market sectors and their impact on the broader market and individual stocks. Two of the most widely-watched sectors are banks and semiconductors – banks because they tend to be interest-rate sensitive and semiconductors because they can be a leading indicator of speculative money entering or leaving the market. Those are simplifications of course, but are among the most common reasons given by traders following those sectors.

The use of this type of analysis is sometimes very helpful – like the evidence from a week and a half ago that showed Rydex traders pulling money out of the semi sector to a degree unlike any other time in the past three years, and which tipped us off that there was excessive pessimism in that group. With Friday’s plunge in Treasury Bonds (and commensurate rise in yields), there have been many questions raised about what may be in store for banking stocks.

When we go back and look at changes in the 30-year Treasury Bond yield and future changes in the Nasdaq Bank Index, there is a correlation between the two. The correlation is relatively weak at -0.13 (out of a scale of 1.0 to -1.0), but with a sample size of 6700, the probability that this is due to chance alone is virtually nonexistent. Since the correlation between rates and banks is negative, the dramatic rise in yields over the past couple of weeks suggests bank stocks may have some difficulty making headway over the next few months.

The table below shows how the Nasdaq Bank Index performed the given number of days after large rises and declines in Bond yields. By “large”, I mean anything more than a 5.4% increase or less than a -5.4% decrease in yields over a 10-day period. Anything outside of these levels would be more than 2 standard deviations from the long-term mean and could be considered unusual. We’ve seen right about that amount of change in Bond yields over the past two weeks, so this study applies to our current situation.

Rates and Banks Performance of the NDX Bank Index After Large Rises and Declines in Bond Yields 1977 - 2004
	3 Days Later	10 Days Later	30 Days Later	60 Days Later	90 Days Later	126 Days Later	252 Days Later
When Rates Show a Large RISE Over the Past 10 Days…
Avg Return	-0.1%	0.0%	-0.3%	1.8%	4.8%	6.2%	10.8%
% Positive	48%	57%	66%	69%	66%	79%	78%
When Rates Show a Large DECLINE Over the Past 10 Days…
Avg Return	0.3%	1.6%	3.8%	7.7%	9.6%	11.7%	23.6%
% Positive	72%	80%	83%	88%	83%	85%	97%
Average Performance During This Time…
Avg Return	0.2%	0.6%	1.7%	3.5%	5.4%	7.7%	16.1%
% Positive	60%	63%	67%	71%	72%	74%	78%

From the table we can see that when yields have increased more than 5.4% over the past 10 days, the Bank Index was higher 3 days later 48% of the time, with an average loss of 0.1%. However, when yields decreased by that amount or more, then the Bank Index was higher 72% of the time. This type of difference is consistent even out to a year later, though the long-term positive bias to the index over the period studied evens out the results the longer out we look.

It’s clear that large, sudden drops in rates have been better for banks than large, sudden rises in rates. We can also see from the table that drops in rates have lead to better-than-average performance while rises in rates have had the opposite effect, though not to the same degree. While rises in rates have lead to poorer-than-average performance going forward, especially in the short-term, it was not radically below average. The biggest difference was within 10 days after the sudden surge in yields – after that, the Bank Index continued to show a lower average return than a random period, but the percentage of time that it was positive was right about on par with any other time. From the time period studied, it appears as though the recent spike in rates may be a negative for banks over the next couple of weeks, but after that it would be difficult to draw any solid conclusions.

Bottom Line: Watching for accumulation and distribution days is most likely a fruitless exercise with little or no future predictive power.

One of the common nuggets of “wisdom” in the market is that really good days, where the market opens near its low and closes near its high, closes positively and has higher volume than the day before, are a sign of bull markets while the opposite is true of bear markets. Like most of these adages, there is a grain of truth to it (but just a grain).

The common definition of an “accumulation” day is one that closes higher than the day before, and on increased volume. A “distribution” day would be the opposite – a day that closes lower than the previous close and on higher volume. In order to further refine the definitions, for an accumulation day let’s also require that the day open within 0.3% of the low, and close within 0.3% of the high. For a distribution day, the figures are reversed – the futures must open within 0.3% of the high, and close within 0.3% of the low. Anytime we look at how the market opens and closes compared to the high and low for the day, it’s best to use the S&P 500 futures as opposed to other cash-based indexes because it has a “true” open, more reflective of actual market conditions, as opposed to the S&P 500 cash index, which rarely shows any type of a gap open. So for purposes of looking at these accumulation and distribution days, we are going to use data for the S&P 500 futures from 1986 through the present.

Using these criteria, in the bull market from 1997 through 1999, there were 53 days showing accumulation, and 26 days showing distribution. In the bear of 2000 through 2002, there were 43 accumulation days and 34 distribution days. There is a difference there, but it’s certainly not huge. If we look at how the market performed after seeing multiple accumulation or distribution days over the period of a few weeks, it becomes difficult to draw any conclusions at all.

As an example, as of March 22^nd, we had seen only 1 accumulation day over the prior 21 trading days, but we had seen 5 distribution days. That dismal performance should have meant that the market was heading lower, but instead we’ve done nothing but rally since then (of course, the jury is still out about the intermediate-term). Since the 22^nd, we’ve seen 3 accumulation days and no distribution days. Looking at the past 21 days, we’re even at 4 accumulation and 4 distribution days. When we look at the “net” number of accumulation days, meaning the number of accumulation days minus the number of distribution days over the past month, we should see the market perform much better after lots of accumulation days than after lots of distribution days if the common wisdom is correct.

In actuality, this isn’t necessarily the case – in fact, very often periods of excessive distribution days have outperformed periods of excessive accumulation days. The chart below shows the net number of accumulation minus distribution days seen over the prior 21 days since 2000.

While there were times this data acted like it “should”, many times it functioned in the opposite manner – more like an overbought/oversold type of indicator. Look at January 2002 as an example. At that time we had seen five more accumulation days than distribution days in the prior month, surely a positive sign according to market logic. But instead it pinpointed the top of the move.

The next time you hear someone suggest that the market is in “accumulation” mode because it opens and closes well on good volume, you should be skeptical. It is much more likely that they are simply repeating something they heard somewhere once than it is that they have hard evidence to back up their claims. When we look at the data, it becomes clear that accumulation and distribution days are a fuzzy concept that should not be relied upon to determine whether the market is more likely to go up or down in the future.

Bottom Line: As with most major holidays, market performance is consistently positive the day before Good Friday and negative after.

With Good Friday this week, it’s time to show the historical performance of the market around the holiday. Like most of the other major holidays, the market has performed consistently – up before the holiday, and down afterwards. The graph below shows how the S&P has reacted the two days before Good Friday and the three days after traders return since 1950.

The day before the holiday, the S&P has closed positively 67% of the time with an average return of 0.3%. The day after, however, the S&P was higher only 37% of the time, and the average return dropped to a negative 0.2%.

If we look at those times the S&P was trading above or below its 50-day moving average, then its performance becomes more clear.

When the S&P was above its 50-day m.a., then the day before Good Friday was higher 76% of the time. But if the index was below its moving average, then the day after the holiday was higher only 10% of the time (2 out of 21 occurrences), with an average return of negative 0.8%.

Near the end of February, I showed the growth of $10,000 if you had went long the S&P 500 for the last two days in February and the first three days in March. The results were very positive, and it worked a bit this year as well, as you would have ended up with a gain of 0.6% over that time period. Using the same concept, the chart below shows the growth of $10,000 by going long the S&P on the close two days before Good Friday, then selling your long and going short on the close the day before the holiday, then covering your short on the close the day after the holiday. This is simply a way to see how the trades would have gone by being long the day before the holiday, and short the day after.

We can see that the growth of capital is very consistent and with minimal drawdowns. The $10,000 would have grown into just under $13,100, all while being in the market for only 109 days. That translates into a theoretical annualized return of over 70%. I would never suggest basing a trade solely on seasonal tendencies, but I do think it behooves everyone, especially short-term traders, to be aware of this strong bias around Good Friday.

The action over the past week has been entirely consistent with an intermediate-term low. While most of our sentiment measures were nowhere near pessimistic extremes a couple of weeks ago, the numerous breadth studies I showed suggested an intermediate-term low was imminent. With the lack of support from our sentiment measures, I wanted to see some confirmation of strength (such as a rally and hold above 1125 on the S&P), and we’ve got it. Our short-term sentiment measures have been rendered useless, as the market has gained steadily in the face of some severe overbought short-term conditions. Last week, I showed some stats suggesting that if the gap in the S&P futures from March 25^th was not closed within 6 trading days, then history suggested that it would not be closed at any point 10 days later either. This is all classic behavior when coming off of an intermediate-term low, and it is further evidence that we have seen the low for quite some time.

As of the close today, we are still grossly overbought on a short-term time frame, and are now becoming overbought on a longer-term basis as well. The Down Pressure reading for the NDX is now the second-lowest in its history (narrowly beat by 11/14/02), the TRIN is extremely overbought, the volatility measures are stretched from their moving averages, and the put/call ratios have dropped dramatically from the readings a couple of weeks ago. In addition, our longer-term up issues, up volume and cumulative TICK measures are now stretched to a point that has usually indicated a short-term pullback in the offing, even during the rally since last March. I don’t have much of a desire to try to go short in the face of these readings, but I think it would be prudent to wait before trying to establish long 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.