sentimenTrader.com

All rights reserved.

Sunday, September 8, 2002

First off, let me get some quick administrative notes out of the way. After a hellish week of the site and my email being in various stages of disrepair, I believe everything is now up and working properly other than inconsistency in the password protection needed to enter the site. I expect to have that corrected within 24 hours.  I have changed my email to a dedicated email host, so capacity issues (such as the dropped or delayed emails that we saw this week) should no longer present a problem.  Although it has been a pain, the site migration allows new features such as the Indicator Search and Sort section, which is a powerful addition that I believe will be of great use to you over time. There's also a password and subscription expiration date lookup page on the main sentimentrader.com site.  These are just a couple of the interactive features that will become more and more a part of the site, allowing you more control over how and what information is presented.

Last week, I said that I expected several triple-digit point moves in the DOW - in both directions - as well as a pickup in volume (the average post-Labor Day increase has been about 10%).  We actually saw a triple-digit move every day, and volume was about 15% higher than last week.  This week should see less choppy action, and may be very important for the intermediate-tem health of the market.

Indicator Summary:

BULLISH

STEM - This model has spent the entire week in various stages of oversold, ending Friday just mildly so with a reading of 78%.  At this point, if we get even a mild decline with large negative TICK and high put/call readings, the other components are extended enough that the model could quite quickly reach "maximum" oversold.

COT - The 6% decline in the S&P during the most recently released reporting period was enough to spur the most positive action in the futures since the July low.  The commercial traders used the occasion to lighten their net short position by around 6,300 contracts while the small speculators reduced their net long position by approximately 2,600 contracts.  This was enough to move the commercial one-year stochastic to 99, which is the most positive reading since around the September 2001 low.  Obviously, then, the three-month stochastic stayed at 100.  The small spec one-year and three-month stochastics dropped a few notches to 46 and 21, respectively.  This action has helped to improve the Futures Balance Matrix on both time frames, which continue to present a positive intermediate-term backdrop.  I refer you back to last week's commentary in which I compared charts of the current stochastic situation with those of the April and September 2001 lows.  Both of those previous times, the commercials began to increase their short positions while the small specs increased their longs AFTER the sprint off the lows ran about 15%.  This time, even though we rallied a like amount off of the July lows, we continue to see positive developments in these futures positions.  This is unlike most of the large rallies off a low point we've seen in the past 10 years.  The one caveat to reading too much into this is the fact that these positions can change dramatically in the course of a week, so it bears careful watching.

SENTIMENT SURVEYS - Just as last week's numbers saw a very large increase in bullishness, this week's saw an even bigger move back to bearishness.  Every major survey segment shed some of their bullish opinions, with the AAII survey swinging from a bullish ratio (bull / (bulls + bears)) of 63% last week to 39% this week.  The percentage of respondents who were bearish soared to 50%.  You can see from the Indicator Search and Sort section of the site that in the past decade there have only been 20 other readings of AAII Bears at or over 50%, and 14 of those were in 1990 alone (to find this information, simply go to the Indicator S&S section, click on [AAII Bears], then enter a "low" reading of 50 and a "high" reading of 100 - this will give you all occurrences of bears being between 50% and 100%).  This is a particularly noisy survey, and I give it less weight than the others (especially individual weekly readings).  Other than that survey, there were no other major changes, although as a whole the increase in bearishness was enough to push the AIM model back to positive territory.

AIM - After dropping 29% last week, this week the model gained back over 30%, putting it back above the bullish threshold of 80%.  Such large increases in bearishness are rare, as there have only been 13 larger increases in the past 15 years.  Here is the performance in the S&P after such jumps:

We can see that such a bearish surge has typically lead to short-term outperformance, as the average return and percentage of time the market was up went up dramatically.  After one week, the market was up an average of 0.43%, with a "win" rate of 62% (compared to random readings of 0.20% and 56% respectively).  After two weeks, the average return increased to 1.17%, with the market being up 77% of the time (compared to random readings of 0.39% and 59% respectively).  After two weeks, the jump loses any edge it may have garnered, and in fact in longer time periods it actually lead to UNDERperformance.  This reflects typical market behavior of "dead cat" bounces after large declines, which ultimately fail and lead to lower lows.

COMPOSITE MODEL - I said last week that any further selling could push this model into positive territory, and that's what happened.  Coming into the new week, it would be quite easy for the model to reach strong positive territory, as the improved sentiment survey and commitments of traders data will help improve this model's score.  If we get some downside accompanied by a few high put/call readings, negative breadth and a spike in the VIX, that would likely be enough to make this model hit 80%, which I would consider to be very significant.  Obviously, as we saw in late July, just because the model is high (and sentiment is terribly negative), it does not mean the market has to stop the decline (even though the odds say it will soon), but it puts us on alert that when price does change, the move will be significant.

PUT/CALL RATIOS -  We have seen a gradual increase in public put-buying, enough to push the 10-day moving average of the total put/call ratio close to bullish territory at .86.  Since 1995, when the 10-day total p/c ratio PEAKS at .86 or higher, this has been the average S&P 500 performance:

I only considered instances where the ratio peaked at or above this level, not every day the ratio was this high.  By peak, I considered occurrences where the ratio reached .86 or higher, then declined the next day.  So in order for us to consider these statistics in our current situation, we must wait for the 10-day p/c ratio to turn down.  It will take a reading under .80 early next week for it to do so.  When it does, we can see from the table above that such occurrences have been very positive for the market in the short-term, with both an average gain and "win" rate much greater than random.  This is NOT due to the market being in a bull market for the majority of the study, as half of the instances came after the market peaked in 2000.  Meantime, the OEX put/call ratio continues to hang around its lower levels, suggesting the bigger-money group of traders are not so bearish on near-term market prospects.  In fact, the current 10-day OEX p/c ratio is in the lower 30% of its 8-year range, while the 10-day total p/c ratio is in the upper 95%.  This tells me that the small equity options traders (who are most often wrong at market highs and lows) are significantly more bearish than index options traders (who are often right at highs and lows).  In the S&P 500 options, there does not seem to be any specific bias, as the bid/ask bias ratio has remained around 1.0 for the past few days.  Volume has been quite heavy, but there is no discernable directional preference.

TRIN - I suggest you read the study posted to the Research section of the site on 6/9/02 regarding high 10-day TRIN readings.  The study looked at instances where the 10-day TRIN peaked at or above 1.65 since 1966.  The study concludes that such readings have historically been extremely positive for the market, with average returns and percentage of time the market was higher from 1-60 days out much greater than random.  The current 10-day TRIN is right at that threshold, so we need to wait for a reversal in the average before we could consider this research applicable to our current situation.  As is obvious from the chart on the site, the recent past instances of such an oversold TRIN have not been very effective more than a few days out.  I've been saying for several months that the oversold TRIN readings have not been nearly as effective as the overbought readings, and that makes perfect sense given we are in a bear market.  The last time the TRIN became overbought should be fresh in our memories, as it was just a couple of weeks ago and lead to the decline we're still struggling with.

BEARISH

NONE

NEUTRAL

BREADTH RATIOS - In late August, I began mentioned the one-year stochastic reading of the 10-day advance/decline line, and when it reached 90% or so, that it would be wise to tighten stops on any long positions and wait for the market to work off its overbought nature.  We did reach 90% near the recent top, and we have since worked off that overbought reading to a great degree.  At this point, the one-year stochastic is at 58%, which basically tells us nothing, and it's likely to stay that way unless we get a sustained move from these levels.  The past two weeks have seen alternating days of large positive and large negative a/d readings, which makes it hard for this indicator to go much of anywhere.  The daily cumulative TICK indicators have also worked off their overbought conditions and are well on their way to becoming oversold.  Before they get become oversold, however, the intraday indicators will have to work off their overbought readings (particularly on the NYSE).  The NYSE intraday indicator is now close to "maximum" overbought (I put the maximum in quotes because the theoretical maximum is 39,000 but the practical maximum is around 7,000 - we very rarely see readings higher than that).  Since the NYSE appears to be in a 30-minute downtrend, this overbought reading is troubling for the very short-term.  Over the past two years, when the NYSE intraday cumulative TICK indicator is as high as it is now, and price is in a downtrend (simply defined as being below the 130-period moving average on 30-minute bars), the average return 3 hours later is -0.27%, with the market being positive only 23% of the time.  The longer the TICKS stay high while price is in a downtrend, the more severe the eventual drop usually is.

SEASONALITY - I mentioned last week that seasonality influences would suggest that there would be a negative bias going into the week, and that played out this year as well.  That's what all of these statistics are for - determining what inherent influences are impacting the market at any given time.  They are not a trading system unto themselves, only a general guide for determining what is most likely to happen during a given period of time.  September has historically been the worst month out of the twelve for equities, on both an average return and percentage positive basis, and by a wide margin.  The coming week doesn't seem to have any particular bias, as the daily average returns and percentage of time the market is up is close to random each day (from the Daily Market Bias section of the site).  So other than the general September downward bias, there does not seem to be any seasonality influences for the coming week.

VIX and VXN - On Monday, the VIX spiked high enough to stretch it more than 20% away from its 10-day moving average, which has historically been positive for the short- and intermediate-term performance of the market, although it has also preceded some dramatic short-term selloffs.  That's why I continually say that although such large volatility movements are historically positive, it also puts me on guard for quick and sharp adverse moves.  You can look at this same information yourself on the Indicator Search and Sort section of the site.  Currently, we are in our 7th day of the VIX being elevated above its moving average (I consider elevated to be more than 5% above the 10-day moving average).  These stretches of the VIX being elevated are even more positive for the market than the fact that the VIX is stretched in the first place - it shows that not only is fear high, it is also persistent.  That combination is very conducive to a higher market in the short-term.  To see just how positive these stretches can be, read the 6/13/02 posting in the Research section about occurrences of the VIX being elevated for long periods of time.  The study looked at stretches of 10 days or more, so we have a few days to go before we could consider those stats valid for our current situation.

NYSE MEMBERS REPORT - I said two weeks ago (the August 25th commentary in the archives) that the Specialist Short Ratio for this week should be around 42%-43%, if the correlation between market performance and this indicator held up.  It seems it did, as the most recent short ratio is 43%.  This is neutral, but shows that the public decreased their shorting activity in relation to the specialists near the market high in late August, which is the typical pattern of the public being wrong at high and low points.  Although this report is released with a two-week delay, knowing that there is a correlation between percentage moves in the NYSE Composite Index and the SSR gives us something of a head start, and we can now assume that the ratio will be lower (a positive development) in two weeks when the numbers for the current period are released.  I suspect that in two weeks, the SSR will very likely be at or just under 40%.

STEM.MR  - Friday's rally completely alleviated Thursday's strong positive reading, and this model is now back to neutral.  Unfortunately, many of you did not receive the model changes this week, as I had terrible email problems (which should now be taken care of).  Since the VIX is still elevated, it is difficult for the model to become overbought.  It will take a significant drop in the VIX (and rise in the market) before we will see a negative indication here.

We are in something of an awkward situation this week, as the overbought condition from a couple of weeks ago is gone, but we are not oversold enough to make a solid bet on higher prices.  That makes me net neutral for the moment, but bullish as long as we can hold above last week's lows, which I believe will be an important line in the sand.  If those lows break, I suspect we will visit the July lows in a hurry.  In the meantime, I will be looking at the 10-day p/c ratio and 10-day TRIN, as reversals in those indicators have historically lead to at least some short-tem upside on a consistent basis.  I wouldn't be comfortable holding anything other than short-tem positions at the moment, with tight stops either way.

 - Jason Goepfert