THAT is What We Needed

Bottom Line: A popular breadth indicator has gone from the 2^nd-most oversold reading in 64 years to the most overbought one. Previous overbought readings have resulted in very positive markets a year out, but we may not be seeing the extremes we think we are.

By far, the most common question I’ve received over the past couple of days is regarding the McClellan Oscillator. The oscillator is very simple, and is calculated by subtracting a 39-day moving average of the number of advancing stocks minus the number of declining stocks from a 19-day moving average of the same figure. It is basically a “de-trended” look at market breadth, and normally follows our other straightforward breadth indicators very closely.

I pointed out the historic oversold reading from this indicator on May 10^th, noting that it was the 2^nd-most oversold in the past 64 years. Now, we have another interesting situation with the indicator, in that as of today it is the most overbought in the past 64 years. Interestingly, two of the other extremely overbought readings came soon after a couple of the most oversold readings in history, those being in 2002 and 1998.

As I said, today’s reading of 287 is a new all-time record. There have been 25 other readings since 1940 that have exceeded 200, and I have outlined the S&P 500’s performance after those occurrences in the table below.

Extreeeeemely Overbought McClellan Oscillator readings of 200 and above S&P 500, 1940 - 2004
	5 Days	10 Days	30 Days	60 Days	90 Days	120 Days	250 Days
Avg Return	-1.3%	-0.2%	0.2%	5.6%	8.1%	13.6%	15.9%
% Positive	28%	64%	68%	72%	76%	76%	91%

As we can see from the table, the market didn’t do so well in the short-term. Five days after the Oscillator hit at least 200, the S&P 500 was positive only 7 out of 25 times, and the average return was a pathetic -1.3%. However, within 10 days the overbought condition wore off, and the market began to recover in a majority of cases. Within 90 trading days, the S&P was positive 3 out of 4 times, and the average return sprinted to more than 8%. After one year, 20 out of 22 cases were positive, and the average return had climbed to a very respectable 16%.

This confirms the typical market behavior when coming out of deeply oversold conditions. Normally the extreme oversold nature of the market indicates washout-type conditions, and the market very often rebounds back strongly in the days after the low, usually becoming extremely overbought within a few days. While the market normally takes a breather for a few days after that overbought condition is achieved, when it wears off a bit we usually resume the uptrend off the low.

The problem with this indicator, though, is that it does not adapt to the number of issues being traded at the time. So of course an Oscillator reading of +200 is easier to achieve today than 60 years ago because there are a couple of thousand more issues trading. If we make a very simple adjustment to the indicator to account for the number of issues traded, today’s reading basically falls off the “extreme” scale, as it ranks only 244^th since 1940. Looking at all readings as (or more) extreme than our current one using this adjusted data, the market basically followed the same pattern – slight weakness in the short-term, then strength thereafter. However, since there were many more occurrences, the performance was watered down significantly. One year later, the S&P was higher 79% of the time, with an average return of 12%. This is still a bit better than an average one-year return during the study period, but it is significantly less than the extremely positive results noted in the table above.

Bottom Line: For the fifth time this month, put/call ratios have been exceedingly high on days the market closes higher than the day before. We have to go back nearly a decade to see similar activity, which lead to a continuation of the uptrend already in place.

I have my misgivings about reading too much into the headline put/call ratios, as I’ve discussed many times, but the data from the Chicago Board Options Exchange (CBOE) continues to be historically interesting. For the fifth time this month, the total put/call ratio from that exchange has shown more puts traded than calls while the S&P 500 closed up for the day. While I believe it’s too simplistic to say that this shows massive betting against rally attempts, it is still very rare. In fact, going back to 1989, I can find only one other period in time where there were so many days clustered within a month of each other when the put/call ratio closed above 1.0 while the market rose on the day – March/April 1995. The chart below shows the five days (the dotted lines) where the p/c ratio closed above 1.0 on a day the S&P closed higher.

Why this phenomenon appeared when it did is beyond me. We were well off the 1994 low and into an established uptrend at the time, unlike this time around when the market has been struggling to gain a foothold. If we look at all 59 instances of the ratio closing above 1.0 on a day the market rose, then the results are not noticeably different than a random time period except after 30 days and beyond. After 30 days, the S&P showed an average return of 2.4% (about double the average 30-day return) and 78% of the time the S&P was positive. After 60 days, the average return climbed to 4.4% (again, about double the average 60-day return) and 85% of the instances were positive. Looking at just the past 10 years, the average return climbed to 5.5%, but most remarkably the market was higher after 33 of the 34 instances (the sole loser occurred on 01/30/02).

One of the reasons for the recent high put/call ratios, of course, is due to options on the Nasdaq 100 tracking stock, QQQ. By subtracting out those institutionally-dominated contracts, we can see that the put/call ratio didn’t exceed 1.0 even once this month. The closest we came was a reading of 0.83 on May 21^st. Still, on a moving average basis that ratio is relatively high, as the following chart shows.

This ratio peaked above what was seen in March of this year, and was higher than at any other time in the past year, though we never did approach the “panic” levels of the major bear-market lows. The green and red lines on the chart are six-month Bollinger Bands, a relative measure of extremes, as it is shown each day on the site.

Bottom Line: The seasonal bias around Memorial Day is weak compared to other major holidays, and may have changed recently. It is becoming increasingly difficult to “game” market reactions around this holiday.

As per usual when approaching holidays, I would like to reiterate the seasonal market tendencies around Memorial Day. This is not a particularly reliable holiday as far as biases go, as the table below demonstrates. I have used data only since 1971, when it was made an official federal holiday and declared as the last Monday in May. The S&P 500 is used as a proxy for "the market".

Memorial Day Bias S&P 500, 1971 - 2002
	2 Days Prior	Day Prior	Day After	2 Days After	Random
Avg. Return	0.02%	0.15%	0.21%	0.02%	0.03%
% Positive	47%	59%	38%	59%	52%
Maximum	2.9%	1.6%	3.5%	1.9%
Minimum	(1.8%)	(1.2%)	(1.5%)	(1.7%)

We can see that Memorial Day follows a fairly common pattern among holidays - the market is often up the day preceding the holiday, and down the day after. 59% of the days preceding the holiday were up during this time period, as opposed to only 38% of the days immediately after. This hasn't been an especially reliable pattern lately, as four out of the last six years have seen the trading day before Memorial Day end negatively. Historically, if the day before Memorial Day ended in the red, then there was a 54% chance that the day after the holiday would also be down. However, if the day before the holiday closed positively, then there was only a 32% chance that the day following the holiday would also close positively. These aren't particularly strong patterns, and the sample size is relatively small, so I certainly wouldn't place trades based solely on this information.

The table above does not include the numbers from last year. In 2003, we saw another instance of the market performing opposite its historical bias. The day before the holiday, the market was mixed (S&P up slightly, NDX down slightly), but the day after the holiday both indexes exploded higher (S&P +2%, NDX +4%). I suspect we could be in for a similar experience this time around, as the heightened talk of terrorist attacks around major events could spook traders in the days before the weekend, but if nothing out of the ordinary occurs, we could see “relief” buying pressure.

Regardless of how the market performs, I would like to personally thank each and every one among you who have sacrificed your time, energy and loved ones to help keep our nation free. We are able to do what we do because of the magnitude of your efforts!

In the last comment, I noted that based on work with how the market has typically responded off of extreme oversold conditions, it appeared as though we had more work to do on the downside unless we could get going NOW. “Now” we did, as the broader indexes put in very solid performances yesterday, and that helped to confirm what the research showed last week. Most of what I have discussed the past two weeks has suggested we have seen a low of some import, and I continue to approach it that way, particularly given the good action the past two days.

In the short-term, as you can see from our intraday indicators, we became extremely stretched by yesterday’s close. The indicator “score” for our shortest-term measures became maximum overbought (for all practical purposes), and the STEM.MR model hit its lowest level in over six months. As I stated in the intraday comments, this type of activity can clue us in as to the market’s prospects for the intermediate-term. If the broader market can hold in here, either by gaining ground or losing just a little, it provides us with even more confirmation that we have seen the lows for some time. Markets emerging from intermediate-term lows often become extremely overbought right off the low, and work off the condition by moving sideways or higher. Lately these types of readings have resulted in lower prices almost immediately, but already we’re seeing some resistance to that pattern. Watch for these indicators to relieve some of their overbought conditions, and if the market begins to take off again, there should be some good gains to be had on the long side.

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.