sentimenTrader.com

All rights reserved.

Sunday, October 20, 2002

I stated last week that from a sentiment perspective, the risk/reward was still heavily skewed to the long side, as any potential downside was outweighed by the potential for further gains.  I also suggested short-term traders look at strength as an opportunity to exit long positions and begin to look for short candidates.  That was mediocre advice unless you were nimble enough to catch the large gaps down that occurred twice.

This week's gap phenomenon was something to behold, and if recent history is a guide, it does not augur well.  In the past five years, there have been 11 instances of 4 gaps of 2% or larger within 5 days (not including this past week).  The average S&P return 30 days later was minus 5.09% with a probability of the market being higher of only 27%.  After 60 days, the average return was minus 6.67%, with a probability of being higher of 27% again.  These statistics are somewhat misleading, however, since there were really only three distinct time periods that saw such gaps - mid-November 2000, mid-March 2001 and early-May 2001.  Due to the small sample size, we cannot draw any solid conclusions about these types of occurrences, but it should be noted that each portended a significant market movement over the intermediate-term.

Indicator Summary:

BULLISH

COMPOSITE MODEL - After becoming extremely positive at last week's low, this model has cooled off dramatically.  The current reading of 59% is down 22% from last week's levels, and significantly below the upper standard deviation band of 77%.  While not at all bearish for this model's time frame, it does suggest that the best long entry point from a risk/reward perspective is behind us, and any long entries taken hereafter should probably have a shorter time frame with tighter reins (stops) than a position taken last week.  Although the longer-term components of the model (e.g. commitments of traders positions and sentiment surveys) will remain positive next week, several of the shorter-term components are quite negative.  The VIX deviation from the 10-day average and 10-day TRIN are overbought, and the 10-day advance/decline oscillator will also likely become overbought this week.  This battle between time frames should keep the model relatively muted this coming week, although it could spike higher should we see a selloff in equities.

SENTIMENT SURVEYS - The spread of nearly 15 percentage points between the Investor's Intelligence bears (43.2%) and bulls (28.4%) is one of the wider in the past twenty years, and hasn't been exceeded in over seven and a half years.  Each of the other occurrences have tended to occur not in isolation, but as a string of readings lasting several months.  I've pointed out the 1994 period several times, but there were also wide bear/bull spreads for several months in 1988 and 1990-91.  If you are looking for a bear-market bottom, and not simply an intermediate-term rally, then you should expect to see the spread stay this wide for several months, and not get excited just that the bulls and bears have reversed their usual position for a couple of weeks.  The other major surveys didn't show any movement of interest, although there was a pickup in bullishness in the noisy Market Vane and AAII surveys.  The longer-term moving averages of these surveys remain positive, however.  The 30%+ rally in the German Neuer Market over the past week stirred up a lot of bullish feelings in the small private traders in that market.  While the institutional traders were bullish coming into the week, the private traders were nearly the most bearish they had ever been.  The large rally swelled the percentage of bullish private traders to 52% (from 24% last week) and caused a drop in bearish traders to 30% (from 51% last week).  Since this market shows a close correlation to the NDX here in the U.S., it adds a note of caution to this complex of indicators.

AIM - The rise in bullishness in many of the surveys offset the large drop in bullishness in the II poll, thus causing this model to remain relatively unchanged.  At a current reading of 65.6%, it remains above the upper standard deviation band of 63.2%, so this longer-term indicator is still strongly positive.

COMMITMENTS OF TRADERS - Continuing the string of relatively unchanged indicators, the commercial traders in the S&P 500 futures pits increased their shorting activity to a slightly greater degree than their long purchases, making for an increase in their net short position of approximately 1,600 contracts.  The small specs increased their purchases and short sales about the same amount, for a change in their net long position of +317 contracts.  These movements have decreased the bullishness of the three-month and one-year stochastics for both groups, but not by a notable amount.  There was a significant change in the dynamics of the Nasdaq 100 futures, with the commercials getting cutting their net short position in half while the small specs went from net long 5,000 contracts to net short 2,300 contracts.  To be honest, however, I have not found a good correlation between these traders' positions and future market movements.  I went over the COT information for 10-year Treasury notes last week, and suggested that we keep an eye on it, since it may provide a clue as to the longer-term direction of the 10-year yield and - by association - equities.  This week the commercials added to their record net-long position in the 10-year futures, by a combination of adding long contracts and reducing their shorts, suggesting they are expecting the yield to fall (and prices rise), which would likely correspond to weakness in equities.

BEARISH

RYDEX RATIOS - Similar to last week, it's important to clarify time frames here first.  In the intermediate-term, the ratios are neutral, after reaching historic levels of bullishness at last week's low (see the bullish and bearish deviations and the asset flow composite chart on the site).  Shorter-term, however, the complex is quite bearish.  The enthusiasm with which this group of traders has made its way out of the bearish funds and into the bullish funds has caused the 3-month bullish ratio stochastic (which compares the current bullish ratio - a comparison of the flow into the bullish funds versus the bearish funds, weighted by leverage - to all of the other ratios over the previous three months) to push above .60, which is the highest it has been since late August.  Also, the 3-day RSI spread, which measures the short-term momentum in the two groups, reached +83 on Thursday.  The chart below shows the performance of the S&P 500 in the days following such readings over the past two years, compared to a random return.  The percentages below the bars are the probability that the S&P 500 was positive the given number of days later.

We can see from the chart that the day following +80 readings actually performs better than a random day, but is neutral at best.  After 3 days and 5 days, however, the excessive bullishness begins to take its toll, and the market suffers.  We can see that after 5 days, the market was negative every single time.  There were 8 such occurrences, so the sample size is quite small, but this still suggests caution is in order.

TRIN - I mentioned on Thursday that the 10-day TRIN was becoming overbought, and we are extremely so now - at least in the context of this bear market.  I've been stating for several months now that oversold readings in a downtrend are not particularly effective, but overbought readings are.  You can see this clearly on the charts below.  The top chart is the 10-day TRIN versus the S&P 500 for the current bear market (March 2000 - current).  The green and red bands are 1.5 standard deviations from the one-year mean, and the red arrows show where the 10-day TRIN met or exceeded the lower band, suggesting an overbought condition.  I'm not a real big fan of comparing current situations to those from previous periods, simply because market dynamics change so much.  But breadth ratios have remained fairly reliable throughout the past 60 years or so, so I feel more comfortable comparing those to other time periods.  The bottom chart shows the same data as the top one for the early 70's bear market.  The point is to demonstrate that overbought TRIN readings in the context of a bear market are not good indications of further strength.  Although the market can and does make progress after such overbought readings, it is rare and relatively tentative.

CURRENT BEAR MARKET:

PAST BEAR MARKET:

NEUTRAL

STEM - The persistently low readings in the TRIN, p/c and VIX on Thursday and Friday served to keep a lid on this model, as it hovers near the lower standard deviation band.  Over the past six months in particular, equities have had a difficult time making significant headway when this model was so low relative to its trading bands.  In fact, when the model has been within 10% of its lower band, the maximum three-day move in the S&P has been 2.4%, while the minimum return has been minus 8.4%.  With such a skewed risk/return, this model is urging long-side caution.

STEM.MR - Other than tipping us off to the large gap down opening on Wednesday, this model did a pretty poor job of catching the major moves this week.  Most of the reason is due to the fact that the model has been revised (see last week's commentary), which makes it a bit slower-moving than it was before.  Although that is detrimental on weeks like this, which are extremely unusual, it has bettered the performance of the model in more normal times.  In any event, the model ended the week in neutral territory, with the cumulative TICK offsetting the relatively overbought readings in the TRIN, p/c and VIX.  This model is telling us nothing about probable short-term direction at this point.

SEASONALITY - We had the pre-expiration positive seasonality working for us last week, but this week we have the opposite.  Although October as a whole tends to be positive, the upcoming week has some bearish overtones.  Namely, the days of the week following expiration tend to underperform a random day, in both average return and probability of being positive.  Also, the specific days of October which make up this week have tended to underperform over the past 50-odd years.  From the Daily Market Bias section of the site, we can see that each of the five days that make up next week have tended to be negative, with the 22nd and 25th of the month being the most negative of any of the October days.  Since 1950, the S&P has been positive on October 22nd only 31% of the time (with an average return of minus 0.25%), while the 25th has been positive 33% of the time (with an average return of minus 0.32%). 

PUT/CALL RATIOS - Other than the one-day spike in the equity and total put/call ratio on Wednesday, this week's readings were subdued, especially for an expiration week (i.e. the equity and total p/c ratios show a 7% upward bias during expiration weeks compared to non-expiration weeks).  I find much more value in the moving averages of these indicators than one-day readings, and although the week showed several low readings (close to touching the lower standard deviation bands), the 10-day and 21-day averages are still somewhat elevated and are close to their upper standard deviation bands.  The OEX put/call ratio was fairly flat all week and its averages are still hovering near the upper deviation band.  Since the averages are elevated for the equity, total and OEX p/c ratios, I would rate the complex as a whole as neutral to very slightly bullish.  You can see all of these ratios and their standard deviation bands updated each night on the site.

BREADTH RATIOS - I pointed out last week that we'll be dropping a string of large negative readings from the 10-day advance/decline line, and we're still in the midst of that.  We have three more days of large negative readings being dropped, for a cumulative total of -4327.  This will make it very difficult for this oscillator to become oversold, but quite easy to become overbought.  At 54%, the 10-day average of the up volume ratio is already close to overbought.  In the next three days, we'll be dropping two low readings, which will help to push this indicator higher.  The cumulative TICK indicators are all currently neutral and not giving us much of a clue at the moment.

VIX and VXN - The VIX is now around -10% from its 10-day average, which is a good approximation of this indicator being overbought.  The last time it poked out of this band was the late-August period, which of course was closely followed by weaker prices.  The VIX fear premium, which compares implied volatility (the VIX) against historical volatility (and can be found daily on the site) is still elevated, although it has begun pulling away from its upper standard deviation band.  This is suggesting that we may be overdone on the upside in the short-term, but that there is still considerable apprehension built into the VIX, which indicates the intermediate-term is more positive.

NYSE MEMBERS REPORT - Basically no change here, as the Specialist Short Ratio dropped to 36% from 37% last week.  I would only consider readings under 35% to be significant.  One notable development is that for only the second time in history, the public has shorted more than the NYSE members for 5 weeks in a row.  The only other time was in 1996-97.  However, as I've stated several times, due to many factors including the increasing ease of short-selling that the proliferation of short-oriented hedge funds, the amount of public short selling has shown a secular uptrend in the past 20 years.  While the amount of short selling we're seeing is notable, it is not as bullish as it seems on the surface.

We have a mix of indicators and models this weekend, which is a good sign to stay cautious on either side and not be particularly aggressive, especially for longer-term traders.  In fact, those intermediate-term traders who established positions when the risk/reward was so skewed to the long side last week should tighten your trailing stops considerably, or wait for lower entry points if you're heavily in cash.  For short-term traders, the evidence is stacked heavily in the bearish camp, and I would treat strength as an opportunity to sell and sell short, not buy.  Since March 2000, when the S&P has been up 7 of the past 9 days (as it has now), the ensuing two weeks have been up less than 35% of the time.  Those are not good odds to fight.

 - Jason Goepfert