Jun 23, 2003

I’ve been mentioning the severe nature of many of the sentiment surveys over the past few weeks, with Investor’s Intelligence first becoming extreme, then Market Vane. Now we can count the AAII (American Association of Individual Investors) poll among those. The most recent survey, released today, showed 71% bulls and 9% bears. This gives a bull ratio (bulls / (bulls + bears)) of 89%, the second-highest reading in the history of the survey, which goes back to July 1987. While we’ve seen fewer bears before (three times) and more bears (one time), we haven’t ever seen this wide of a spread between bulls and bears during the same week.

The following table shows what happened after each of the top 10 bull ratio readings since July 1987. The data is actually skewed by a few days since I have recorded the readings as of Fridays and not Wednesdays, but that doesn’t impact these results too terribly much. The table ranks the top 10 bull ratios from the most extreme ratio to the least, with the maximum and minimum returns in the S&P 500 over the next 21 and 42 days (corresponding to the average length of one month and two months, respectively).

		The Next 21 Days		The Next 42 Days
Date	Bull Ratio	Max	Min	Max	Min
08/14/87	92%	1.4%	(7.5%)	1.4%	(9.8%)
06/20/03	89%	?	?	?	?
09/01/00	89%	(0.2%)	(9.0%)	(0.2%)	(17.4%)
09/22/00	88%	0.6%	(11.4%)	0.6%	(11.4%)
11/03/00	88%	1.2%	(9.0%)	1.2%	(12.6%)
12/11/92	86%	2.1%	(2.2%)	3.9%	(2.2%)
05/17/96	86%	2.0%	(1.2%)	2.0%	(9.7%)
12/31/99	85%	1.1%	(7.7%)	1.1%	(9.6%)
12/03/99	85%	4.0%	(2.4%)	4.5%	(4.7%)
07/14/00	85%	0.7%	(5.3%)	2.4%	(5.3%)

AVERAGE		1.4%	(6.2%)	1.9%	(9.2%)

We can see quite clearly that this display of bullishness was typically not healthy for the market. Out of the top 10 occurrences, the maximum return over the next month was 4.0%, while we saw as much as an 11.4% decline. The average rise was 1.4% while the average decline was a large 6.2%. This discrepancy becomes even more pronounced when we look out over the following two months – then the average rise increases only slightly, to 1.9%, while the average decline slips to 9.2%.

Here are the top 10 graphically (number “2” is omitted because it just occurred):

This survey has often been discounted when compared to the others because it is so volatile week-to-week. It turns out there are other reasons why it is perhaps not the most representative of individual investors. The survey covers a one-week period beginning and ending on Wednesdays, with members of AAII submitting their opinion on the course of the stock market over the coming six months via the AAII web site. They can choose to be bullish (expect the market to rise over the next six months), bearish (expect it to fall) or neutral (not much change either way). I spoke with the person who runs the web site for AAII today, to find out precisely who is responding to the survey. According to her, AAII requires members to sign in to the site before submitting their opinion, and member ID’s are tracked so that each ID can respond only once. This prevents members from voting multiple times, which is good. However, and this is astounding to me, out of 140,000 registered members, an average of only 90 to 100 members respond to the sentiment survey each week. This is an incredibly small sample size, given the population of the organization. I asked her if they keep track of which members respond each week, and they do not. So they don’t know for sure if they have the same members taking the poll each time, or if it varies from week to week. My guess is that they have a number of “die-hards” who religiously go to the web site to respond each week, along with new members who are interested in trying out the various features of the site. Given the extremely small sample size and methodology, it is difficult to place a lot of weight on these results. While the survey has been an effective contrary indicator on a moving-average basis, and it should continue to be so, I think it’s important to keep in mind the inner workings of the poll.

For the latest month, short interest in Nasdaq shares increased. However, short interest is a very cyclical phenomenon, with it being lowest early in the year and highest later in the year. Most traders compute a short interest ratio by taking the short interest and dividing it by average volume during the past month. The bad thing about that is that volume is also cyclical. So, the short interest charts as posted to the site have been adjusted to remove the seasonal tendencies from both short interest and volume. When we do that, this adjusted short interest ratio actually fell quite dramatically for trades taken through the month ended June 10^th (the latest information available), as the following chart shows.

We can see that the record level the adjusted short interest ratio reached in March has come down significantly, to a level last seen in June/July 2002. I certainly wouldn’t say this information is yet at a level where it could be considered bearish, but it does show us that there is now less potential “fuel” from short covering.

If we look at the same information for NYSE shares, a similar though more troubling picture emerges.

Here, short interest never got as high as it did on the Nasdaq, and in fact has remained relatively low throughout the bear market, at least compared to the levels we saw during the bull of the late 1990’s. The current reading is the lowest since December 2000, though historically it is still quite high. Obviously, since the data is released monthly, this is a not a precise timing device. But it does give us a general idea of what kind of force there may be behind future rally attempts.

Some of you may question the apparent dichotomy between this information and the heavy level of public shorting on the NYSE that I have been discussing for several weeks now. You’ll recall that public shorting as a percentage of the total, when viewed on a longer time frame, is at one of the most extreme levels in history (see the May 25^th weekly commentary in the archives for some background on this). This should be bullish, since each of the other eight times this type of activity has been encountered since the 1940’s, it coincided with a multi-year move higher in equities. So how can we reconcile the fact that short interest is declining while public shorting remains historically high? The chart below plots the information above (the NYSE short interest ratio) against a 26-week moving average of the public short ratio.

We can see that these two series track quite closely. Recently (meaning over the past year or so), public shorting has remained extremely high while the short interest ratio has come down. This suggests that someone is covering their shorts, but it’s not the public, which should be bullish. During the past couple of months, this has started to change, but there is still a lot of room to go.

On Tuesday, I said that if we had another large down day on Wednesday, then I would look quite aggressively for short-term long plays, as some of our shorter-term measures indicated that that type of selling was likely unsustainable. Today’s reaction worked off many of those oversold extremes, with only our shortest-term Rydex asset flow indicator now giving us oversold readings (though even that could change after today’s numbers are reported later this evening), so short-term our indicators are neutral to slightly overbought.

From a sentiment perspective only, supports for this market on anything other than a very long time frame are rapidly deteriorating. This weekend, I suggested that perhaps the most likely scenario going forward was a wide trading range, lasting long enough to wear out some of this ebullient enthusiasm we’re seeing in many of our indicators but still respecting the incredible thrust we’ve seen off the March lows. I continue to think that’s the best scenario going forward, so as we react back up toward the recent highs, the short side presents a better risk/reward profile with a holding time frame of a month or more. If we are able to exceed the recent highs and hold there, then I would NOT want to remain short, as the potential remains for this market to carry significantly higher before a more serious rest sets in. I greatly miscalculated the strength the market had once it broke the 930/940 level, and I’ve shown several precedents over the past couple of weeks which suggest that even another 10% on the upside has some historical support. So while sentiment in general suggests we look for shorts and hold through the summer, this momentum must be respected.

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.

I’ve mentioned the low level of the equity put/call ratio several times recently. Not only is that ratio quite low (particularly when QQQ options are excluded), but the total put/call ratio is also very low. Due to a subscriber request, I took a look at how the market performed after the CBOE total put/call ratio reached certain relative extremes over the past 8 years. By relative extremes, I mean that I measured the deviation of the 10-day moving average from the 200-day average. So, as options traders scramble for protection by buying put options, the put/call ratio rises and the 10-day average begins to rise above the 200-day average. Conversely, when traders are relatively complacent and prefer to concentrate on call options, the put/call ratio drops and the 10-day average falls below the 200-day average. When the 10-day average gets stretched far from the 200-day average, we can say that sentiment has reached an extreme and we may be near a turning point in the market.

From 1995 through the current period, the mean “deviation” of the 10-day average from the 200-day was 0.0% with a standard deviation of 13%. For the statistical types among you, the skew in this data was fairly muted, so we can consider a +13% deviation equally as extreme as a -13% deviation, which makes bullish and bearish comparisons much more straightforward. This means that when the 10-day put/call ratio stretches more than 13% from the 200-day average, we’re beginning to see something unusual and potentially significant. Currently, the 10-day average is 18% below the 200-day average, so we can say with confidence that we’re seeing an extreme amount of call volume in relation to put volume – normally a sign that the market will struggle going forward.

I was curious to see how long it took for the deviation to “top out”, meaning how many days it took to reach its peak reading before settling back to normal. Over the past 8 years, when traders became fearful and plowed into puts, driving the 10-day put/call ratio more than 13% above the 200-day, it took 7 days on average before the 10-day ratio topped out and began to fall back again and become less extreme (the longest streak was 23 days). It took an average of 16 days – fully three trading weeks – before the 10-day dropped back to a “normal” reading less than 13% above the 200-day average (the longest streak was 47 days). On the other end of the spectrum, when call volume was dominant and the 10-day ratio dropped more than 13% below the 200-day, it took an average of 9 days before the 10-day ratio began to rise again and become less extreme (the longest streak was 33 days), and 18 days before it became “normal” again (the longest streak was 54 days).

We can see from the paragraph above that it took longer for the put/call ratios to bottom out after a string of low readings than it did for it to peak after a series of high readings. This confirms the old adage that “bottoms are events, while tops are processes”. This means that we often see panic-type spike lows where the market bottoms in a few short days, while tops take longer to form as there are repeated attempts to break the market out above previous highs.

As for how the market performed around these put/call readings, the tables below shed some light on that. There are two tables – one for “complacency” and one for “fear”. Complacency is defined as the 10-day put/call ratio falling more than 13% below the 200-day put/call ratio. Fear is defined as the 10-day ratio rising more than 13% above the 200-day. The four general columns across the top show how the S&P 100 (OEX) performed 5 days, 10 days, 20 days and 40 days after each event (“event” is described below), with the average return given along with how often it closed in positive territory. The three rows (the “events”) show three snapshots in time – the first day the 10-day ratio stretched more than 13% from the 200-day; the day the 10-day ratio peaked and started to head back to normal; and the last day it was extreme – after the last day, the 10-day ratio was stretched less than 13% from the 200-day and could now be considered “normal”.

* Selling short when the put/call ratio first became complacent was not a good strategy overall.

* Buying the market when the put/call first became fearful was also not a particularly good strategy.

* Based on this measurement of the put/call ratio, it appears as though the best time to look for a market top is AFTER the ratio goes back to normal, not necessarily after it peaks and certainly not as soon as it becomes extreme.

* Similar to the statement above, the best time to look for a market low does not seem to be immediately after the put/call ratio becomes extremely high. The best time, at least historically, has been after the ratio peaks and begins to come back down.

* Overall, this measurement of the ratio appears to be a bit better at pinpointing market lows than market peaks. However, it should not be a surprise that from 1995 – 1999, the buy signals worked much better than sell signals, and from 2000 -2003, the sell signals were very good while the buy signals were less so. Taking the larger trend into consideration and only taking the contra-signals (i.e. oversold in an uptrend and overbought in a downtrend) worked extremely well.

What can we take away from this that’s applicable to right now? I think we should expect to see this measurement (the deviation of the 10-day put/call ratio from the 200-day) become even more extreme in the following days. If the daily put/call readings stayed about where they are now, the deviation will become even more extreme for the next three days or so. It would take as long as two weeks before the deviation went back to “normal”. Since we can see from the table above that when this measurement gives complacent readings, the best time to sell is after the ratio makes it back to normal territory, this would suggest that the market has another kick higher before it would likely roll over in a bigger way, and we should wait until the deviation makes it back to normal territory before becoming too aggressive on the short side. However, if we weight recent occurrences more heavily than older occurrences, that would suggest the best time to sell is after the ratio peaks, and not necessarily after it makes it back to normal. In any event, all the information taken together tells us that historically, after seeing such complacency, the market has normally taken more time to form a top and has not rolled over immediately.

This weekend, I mentioned that our breadth measurements were beginning to work off that overbought condition they had been in for a couple of months. That continues to be the case, and if we see another down day tomorrow, the 10-day advance/decline line and up volume ratios could actually become somewhat oversold. That would coincide with our Down Pressure readings for the S&P 500 and Nasdaq 100, which are already at or close to oversold now. When they have entered extreme territory, these Down Pressure gauges have done a decent job at giving us a heads-up when short-term breadth becomes too negative, particularly over the past couple of months. The current readings are the most oversold they have been since late March, and another big down day tomorrow would push them into “unsustainable” territory – a good argument that we should see a sizable bounce sometime soon. If we are down big tomorrow, then I would begin to look quite aggressively for short-term long plays.

Many of our shortest-term sentiment measures became oversold by yesterday’s close, although today’s choppy session relieved much of that condition. If we see another down day tomorrow, then there’s a good chance that we’ll see a reaction back up soon thereafter. But with the state most of our intermediate-term measures continue to be in, it does not look as though new highs would be sustainable just yet.

	5 Days		10 Days		20 Days		40 Days
	Avg	% Pos	Avg	% Pos	Avg	% Pos	Avg	% Pos
First	0.8%	63%	0.4%	56%	1.0%	69%	0.9%	63%
Peak	0.7%	63%	-0.2%	44%	-1.9%	38%	-1.3%	44%
Last	-0.4%	44%	-1.7%	31%	-1.9%	31%	0.8%	50%

	5 Days		10 Days		20 Days		40 Days
	Avg	% Pos	Avg	% Pos	Avg	% Pos	Avg	% Pos
First	0.0%	57%	-0.3%	43%	0.4%	57%	0.7%	57%
Peak	1.4%	71%	2.2%	71%	2.0%	71%	5.7%	76%
Last	0.8%	57%	0.3%	67%	1.1%	57%	2.3%	57%