Sunday, June 22, 2003
Last week, I mentioned that data from the Commitments of Traders information for S&P futures showed the most negative change in at least the past six months. This week, things have changed once again, as commercials added 16,500 contracts to their net long position while small specs added a little over 4,000. With a rise in the S&P 500 of about 3% during the reporting week, it continues the pattern of commercial traders going against their historic norm. Usually we see these traders become shorter (or at least less net long) as the market rises, but for the past few months this has not held. Commercials have not been (net) sellers of this rally, at least as far as the full S&P 500 futures contract is concerned. One thing that makes reading this week difficult is the looming expiration, which will be the case next week as well. Late last year, I showed that the position changes the week before and the week of expiration were typically much greater than random, and I think there are so many variables affecting these events that it’s difficult to determine exactly what the underlying trend is.
This obviously affects the e-mini as well, but in that contract commercials have now set a new record net short position, at over 350,000 contracts. Small speculators reduced their net long position by a relatively large amount, but are still long almost 400,000 contracts. I say this every time, but it bears repeating…I don’t believe one can read too much into the e-mini until it establishes more of a history. I have not yet found a way to incorporate the information that is consistently useful, but it is a factor that has not existed before when analyzing the full contract, so I think it should at least remain on our radars.
Similar to how commercial traders in the S&P futures are behaving, specialists on the NYSE have not picked up their shorting activities to any great degree. Specialist shorting, both as a percentage of the total and on a de-trended basis, remains below where it was during any other intermediate-term peak over the past five years even though we’ve experienced one of the most persistent upward trends in price. This is market-positive.
Public shorting, on the other hand, has dropped considerably since the March low, going from 55% at that time to 44% during the most recent release (which covers the week up to June 6th). This is the lowest level of public shorting since December 2001. If we look at public shorting on a de-trended basis, meaning we take the long-term secular trend out of the data, then it becomes a bit more troubling, as the chart below illustrates.
The “de-trending” in this case is accomplished by looking at the four-week average of public shorting as a percentage of the 26-week average. So, it compares action over the past month to that over the past six months. It will oscillate above and below 1.0 as the public increases and decreases its interest in shorting shares on the NYSE, respectively. Currently, the de-trended ratio is at 94%, meaning public shorting since early May has been running about 6% below the level of shorting since the beginning of the year. I have placed red arrows at those times where the ratio formed a low at or below 94%. The only real spectacular failure as a sell signal came in November 1998, when the market continued to rise unabated even after a sharp drop in public shorting and rise in specialist shorting. This was in the immediate aftermath of the LTCM hedge-fund crisis, so that undoubtedly had an impact on these figures. Other than that, the market consistently had difficulty making significant headway after seeing such low public shorting, at least during recent history.
If we go back further, and look at this information during the time most compared to now (the mid-1970’s), then it doesn’t look quite so ominous.
The red horizontal line is at 94%, which is where we are at now. We can see that during this time, public shorting at times became much more extreme, in both directions, than it has recently. Still, generally, high relative levels of public shorting lead to positive market performance while low levels lead to poorer performance. One exception is early 1975, when we were recovering from the bear market in the years prior. You may recall this is also the time we had a similar extremely positive breadth thrust as we have seen recently (see the June 4th daily commentary in the archives).
So, specialists are not shorting aggressively (which is positive), and the public is most certainly not shorting heavily (which is negative, at least in terms of recent history), so who does that leave? The only possibility is other NYSE members. This is indeed true, as “other member” shorting as a percentage of the total is at the highest level in six years. I mentioned several weeks ago that I have never found “other member” shorting particularly useful, and after taking another look this weekend, I came to the same conclusion. Overall, I’ve been saying that this data has been positive, especially on a very long-term time frame. It’s still true that when we look very long-term (out several months to a year or more), the proportion of public vs. specialist shorting we’ve seen has only been matched a handful of times in history, each of which corresponded to a major, multi-year bull move in equities. However, on a shorter time frame (out several weeks at least), I think the low level of public shorting has to begin being a concern. I don’t think the data is yet outright bearish, especially since specialists haven’t been shorting heavily, but this bears careful watching.
The sentiment surveys, for the most part, continue to become more extreme in their respondents’ show of bullishness. The one survey which shows long-term secular trends more than the others is Market Vane. This poll surveys leading CTA’s (Commodity Trading Advisors, or futures traders), and has quite a long history. Typically, the trend of the survey mirrors that of the market. When equities in general are in a long-term bull phase, this survey tends to give higher readings of bullishness. When the market trend is down, so too is the Market Vane bullish percentage. For example, during the bull phase of 1995 through 1999, Market Vane showed a mean bullish reading of 54% (with a standard deviation of 11%). But from 2000 through 2003, the mean dropped all the way down to 34% (with a standard deviation of 9%). Because of this inherent quality of the survey, once again I think it is instructive to view the data in a de-trended manner, by looking at the readings over the past month compared to those over the past six months. When we do this, the current reading of 56% bulls has pushed the one-month average to nearly 50% greater than the six-month average. This is a nearly 3 standard deviation move, meaning that it is extremely unusual historically. In fact, it has only been exceeded once in the past 20 years, that being August 1984. At that time, after seeing such an extreme amount of relative bullishness, the S&P 500 traded in a tight trading range for the next five months before resuming its upward advance. As far as recent history goes, the only other times that have approached this type of extreme bullishness were January 2000 and May 2001, both of which of course preceded short- to intermediate-term declines.
In the June 12th daily comment (see the archives for reference), I laid out the details of each past occurrence of Chartcraft’s Investor’s Intelligence survey showing fewer than 16.5% bears. At the time, I said that the S&P 500 rallied an average of 49 days (and rose an average of 9%) until it suffered a correction of at least 5%. So far, we’ve rallied about 3% over the course of 10 days since the first reading below that threshold, so according to the “average”, we’re about a third of the way along before a more severe correction sets in. I’m not at all a fan of using these types of projections when we have a sample size of only 13 occurrences, so please don’t think the market HAS to rally another 6% over the next 40 days or so before we decline. The point of that study was to show was has happened in the past, and that it was unusual to see the market decline immediately after seeing a low level of bears. In any event, the bears dropped again slightly during the most recent reporting week, which shouldn’t be a surprise. Again, the average length of time before the bears went back above 22% was 10 weeks, so it’s entirely possible to see such low bears for weeks to come.
Many of our shorter-term breadth measurements have finally come back to neutral, after spending the last two months in various stages of overbought. The 10-day averages of the advance/decline line and the up volume ratio have come down to the lowest readings since March or early April, though they are not nearly oversold. There are two ways to take this…either momentum has slowed to an extent that has not been seen during the last stage of this rally (bearish going forward), or the market has finally taken enough of a breather that breadth has been allowed to come in a bit (bullish going forward). I fall somewhere in between those two characterizations, in that I think we’re finally seeing some days of distribution and sloshing around that we hadn’t really seen since March, which indicates the rally may be getting tired. On the other hand, I don’t think it’s likely that we top out without more of a fight, and the first signs of an oversold market are more likely to be bought than not. So while that indicates we may not see much more upside, it also suggests a large negative rollover isn’t terribly likely either.
As I said last Wednesday, we’re seeing a large layer of froth on the market, and it has only begun wearing off. Until more of this speculation dissipates, and we remain below 1008 on the S&P 500, my preferred focus remains on the short side for anything longer than day-trades. But with the historical precedents I’ve laid out over the past few weeks, the most likely course of action appears to be a trading range for the foreseeable future, so my intention is to focus more on oscillator-type indicators instead of those that rely on sustained trends to give signals. If we decline to some widely-watched support levels (950/960 on the S&P seems to be fairly common), I think the chances for a snapback reaction are extremely high, as are the chances we will be rejected somewhere near the recent highs.
- Jason Goepfert
Disclosure: long OEX puts
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.
© 2003 Sundial Capital Research, Inc. All Rights Reserved.