sentimenTrader.com

All rights reserved.

Sunday, September 29, 2002

A pretty decent week this week, at least in the short-term, as the odds suggested upside was probable after the selling early in the week, then the overbought conditions pointed out on Thursday asserted themselves on Friday.  Hopefully most of your were able to keep or make more money than last.

Indicator Summary:

BULLISH

STEM - The rally off of the record readings registered earlier this week, although notable, was not especially dramatic.  The divergence pointed out last week would suggest that the move we saw is only the beginning, and how the model responds to further downside could be very telling.  If we sell off enough to truly challenge the July lows in the S&P, it's quite likely we'll again set a record in this model and the divergence will once again come into play.  As it stands this weekend, we've come off a bit from the highs in the model, and for the next two days we'll be dropping some low readings.  So it's likely that we wouldn't see any real extreme readings here until late Tuesday or Wednesday if we continue to sell off.

SENTIMENT SURVEYS - A lot has been made this week of the Investor's Intelligence survey, and that it actually gained a few bulls during the reporting period.  It sure did.  A whopping 0.3%.  Meanwhile, the bears grew by over 4%.  What's also interesting is that the above statement about the bulls really isn't even true.  The bulls actually lost closer to 5%.  Investor's Intelligence has three categories to which they assign their respondents.  I only report the bulls and bears, since those are the respondents that express a clear opinion.  The other category is Correction, which is defined as bullish but expecting a short-term decline.  The correction camp during the latest reporting period went from 28% to 23.3%, a drop of 4.7%.  So we can see that quite a large number of those who were expecting a short-term decline have now changed to where they are expecting a longer-term or more severe one.  That, to me, does not sound like an increase in bullish opinion.  Each of the other surveys also showed a decrease in bullish opinion, although we are still not close to any real extremes in any of them.  You can see from the stochastics chart on the site that each of them is showing a decrease in bullishness, although they're not near the pessimism seen at the recent past lows.  In what could be considered a bell-ringer of a contrarian signal, Deutsche Boerse AG (Germany's stock-exchange operator) said that it is going to shutter the Neuer Market, which is the basis of the Neuer Market survey posted to the site, and which has a close correlation to the Nasdaq 100 here in the U.S.  After dropping 95% from its highs, the interest in the index isn't what it used to be.  I'm not sure what's going to happen to the sentiment survey, but my guess is they will just shift the focus to include a broader base.  In any event, this week's reading showed the small private traders exhibiting a higher bullish ratio than the institutions, which has only happened on five other weeks this year.  Each of them preceded an acceleration in downward pressure in those shares.

AIM - This model, which is an average of the stochastics mentioned above, ticked up again this week.  The current reading is 1.5 standard deviations from the 10-year mean, and over 1.0 standard deviation from the mean over the past two years.  This is unusually high, but not quite extreme.  You can see from the chart on the site that since this bear market began in 2000, the significant lows have recorded readings over 65%, compared to our current reading of just over 61%.

COMPOSITE MODEL - We continue to churn around in extreme territory here, as many of our intermediate-term indicators remain elevated while the shorter-term ones pop in and out of extreme territory.  At this point, it would take a rather large spike in the VIX and some major retail put buying to get us back up to the highs.

RYDEX RATIOS - Not much to add here from last week, as the 10/50 moving average deviations continue to remain at record levels for both the bull and bear funds.  With the small rally we got this week, the pressure was relieved a bit and the stochastic has begun to tick up from the bottom-scraping levels it reached earlier this week.  If the market continues down, then we will have a situation not encountered before with these ratios.  The stochastic reached similarly low levels in late June, before the cascade lower into July, so that was obviously a failed signal.  However, this time the deviation of the 10-day ma from the 50-day is much, much greater for both the bull and bear funds, so the longer-term backdrop is more supportive than it was in June and even July.  With the relatively short history available, we'll just have to wait to see what impact this has.

COT - This week, we saw another large move for both the commercials and small specs, and the bullish case from this set of indicators is growing quickly.  The commercials decreased their net shorts by 9,700 contracts while the small speculators decreased their net longs by 15,100 contracts.  We have now seen a two-week net positive change of 57,000 contracts (meaning the commercials decreased their net shorts while the small specs decreased their net longs at the same time), which is the third-largest positive move in the history of the data, going back to 1986.  The two larger moves were 63,000 contracts at the low in March 2001 and 70,000 contracts at the low in September 2001.  Of course, with the increased volume and open interest in these futures contracts over the past few years, 57,000 contracts today is not the same at 57,000 contracts a decade ago.  Even when adjusting for contract volume, however, this recent move still ranks in the top 10.  Not surprisingly, it has moved both the one-year and the three-month stochastics for the commercials to 100.  For the small specs, the one-year stochastic dropped to 4 while the three-month is now at 0.  This combination has pushed the one-year Futures Balance Matrix to a very bullish 96%, and the three-month FBM to 100 for the second week in a row.  The one-year stochastic for the commercials has now been 100 for two reporting periods in a row.  This has only happened 8 distinct times in the past 16 years.  Here is the S&P performance after the second week of a 100 stochastic:

PUT/CALL RATIOS -  The equity and total put/call ratios have come off of their extreme expiration-week readings, though they still remain elevated.  In fact, both ratios remained above their long-term and bear-market median values every day of the week except for Thursday, while the OEX put/call ratio remained under its averages every day.  Although not by any means extreme, it is still constructive as it shows that small traders are not piling on with call options every time we get some upside, while the more savvy index traders are not betting a ton of money on a down move.  The 10-day and 21-day moving averages of the equity and total p/c ratios continue to hover significantly above their 200-day averages, although that is likely to change dramatically this week, as we will be dropping a reading at or over 1.0 for both ratios every day.  Even if we get a string of elevated readings of around .90 in the total p/c ratio, by Friday the 10-day average would stand at .86, a decrease of .15.  So unless we put a low in here, the equity and total p/c ratios will have given another false (or at least really early) buy signal, similar to the July low.  One difference this time is the neutral stance by the savvy OEX traders.  Unlike the July low (and almost every other significant low before it), these traders are not betting heavily on upside at this point.  The 5-day and 10-day spread between the OEX and equity averages are low (bullish), but that's because of the week of extremely high readings we will be dropping in the equity ratio, and not because of a confluence of retail pessimism and institutional optimism, which is what I would much prefer to see to have more faith in this complex of indicators.  The true institutions who ply their trade in the SPX options did not show any real strong bias this week, as our put/call bid/ask bias ratio remained around 1.0.  Other than a rather large amount of call selling on Wednesday, this group didn't tip their hand at all that I could see.

SEASONALITY - This week, we get the positive influences of month-end and new-month.  Posted to the site under the "Seasonality" section, you can now see a table showing these biases around month-end for the S&P since 1950.  The last day of a month and the first three days of a new month do exhibit a positive bias.  In fact, these four days are the only days which are better than random for every situation I looked at, including just a normal day, days when the 50-day moving average is sloping up, days when the 50-day average is sloping down, when the day exceeds the prior day's high, and when it exceed the prior day's low.  For Monday, the last day of the month this time around, the bias is especially acute. 

Although nothing spectacular, when we consider that over 10,000 individual days were studied, during every kind of market, the fact that there seems to be any kind of consistent bias at all is rather remarkable.  So for Monday, the fact that it is the last day of the month and the 50-day moving average is sloping down gives us an average return of 0.08% with a 50/50 chance of being positive.  Again, nothing spectacular but it's better than if Monday was just any other day.  Also, there's a 57% chance Friday's high will be taken out (leading to an average closing return of 0.50%) but only a 41% chance Friday's low will be exceeded (leading to an average return of minus 0.39%).  Now of course, considering Friday's action, it's hard to believe the high will be taken out and it wouldn't take much at all for the low to fail, but hey it's fun to imagine, right?  In any event, should Friday's low be taken out, the end-of-month bias still suggests that the downside would be more limited than if it had just been a random day, which is a positive sign.  Also, since the top in March of 2000, the last day of the month has shown an average return of 0.24% (compared to a random return of minus 0.10%), and a probability of being positive of 55% (compared to a random probability of 45%).  So even during this bear market, the end-of-month seasonality has stood.

BREADTH RATIOS - Most of these indicators are in oversold territory, but not extremely so.  Earlier this week, I pointed out the levels of oversold that past lows have reached, so I'm not going to go over that again.  It is HIGHLY unlikely that we will see severely oversold breadth readings again any time soon, as we will be dropping the following advance/decline readings from the 10-day average over the next several days:  -433, -1271, -681, -1765, 161, -1441, -1274.  So with those types of numbers coming off the average over the next week and a half, it would take a severe market pounding to push this indicator lower, and not much of a rally at all to bring it back to overbought.  The cumulative TICK readings are all challenging their bullish oversold thresholds, and this is certainly a positive development.  The intraday indicators have already cycled back down to oversold after doing an absolutely perfect job of giving us a heads-up of overbought conditions within the 30-minute downtrend we were in.  For short-term traders, I suggest you refer back to Thursday's daily commentary and burn that chart of the NYSE cumulative TICK into your memory.  When these indicators reach overbought within a downtrend and oversold within an uptrend, the chances of you having a successful trade in the direction of the trend are enormous.  It's one of the most consistent edges I have found, and I highly suggest you take advantage.  As for now, we are getting oversold within a downtrend, so although it is somewhat positive, these signals fail more than if they were contra-trend.

TRIN - Like the breadth readings above, the TRIN for the NYSE and Nasdaq will be dropping a string of high readings next week, helping to force the 10-day averages lower.  I've been stating for some time now that oversold TRIN readings within a downtrend such as we're in are not nearly as effective as overbought readings (duh!), but when we get severely oversold such as we are now, it does often lead to at some short-term relief.  Following are the returns in the S&P when the 10-day TRIN reaches 1.60 or greater (where we are now) since the top in March of 2000:

We can see that an overwhelming percentage of the time, the S&P significantly outperforms for up to 3 days after such readings are reached, but the downtrend resumes after that a majority of the time.  To be clear, these are not when the TRIN peaks above this threshold, only when it reaches it, so these stats are valid come Monday.

BEARISH

NONE

NEUTRAL

SHORT INTEREST - Our newest entrant to the stable of indicators is the short interest ratio on the NYSE and Nasdaq.  The short interest ratio measures the total number of shares sold short for each stock on the exchange (that is not yet covered) divided by the average daily volume in the securities.  The typical ratio is computed by simply taking the total short interest figures released by the exchanges each month and dividing it by the average daily trading volume during that reporting period.  This is not an ideal situation, because more than any other indicator that I follow, short interest shows a very definitive and consistent seasonality bias.  Most likely due to tax implications, short interest is significantly higher in the later months of the year than it is in the earlier months.  Not only that, but as I have pointed out numerous times (and you've no doubt observed), volume is also quite seasonal.  It is typically quite a bit lower in the summer months than it is around year-end.  So the combination of short interest and volume seasonality biases - the two components of the indicator - can skew this ratio greatly, and the numbers that you get from other sources almost undoubtedly contain this inaccuracy.  To correct for seasonality, I have studied the monthly skew for each year from 1943 - 2002 and normalized the short interest for that bias.  Also, instead of using just one month's volume, I use the past 12 months, which helps to correct for any volume seasonality.  This Modified Short Interest Ratio gives us a much, much clearer read on the actual sentiment behind the short interest numbers.  For the NYSE I have also posted the MSIR (Modified Short Interest Ratio) divided by its 5-year average, to correct for the secular rise in short interest since the bull market began in 1982.  This ratio has a 50-year median value of 0.18 (with a standard deviation of 0.66), and we are currently neutral at 0.09.  When we stretch more than one standard deviation above the median, showing greatly increased shorting activity and pessimism in the market, it has historically lead to positive long-term market performance.  In fact, this was one sentiment indicator that correctly stayed bullish during the incredible bull run beginning in the early 1990s, and became bearish in a historical sense near the top in 2000 as modified short interest declined in spectacular fashion.  Since the bear market began, we have started the long road back to neutral.  Don't read too much into the almost monthly headlines that short interest is hitting a record, since that figure taken alone means squat.  Over on the Nasdaq, where significantly less data is available, short interest is once again reaching quite high levels.  Over the past 8 years, a MSIR for the Nasdaq between 2.5 - 3.0 has lead to good market performance while ratios under 1.75 have not been kind.  With a current reading of 2.51, this indicator is becoming bullish for the shares traded on that exchange.

VIX and VXN - On Thursday, I pointed out that the VIX.MR indicator reached the 20% level, which had coincided with recent past highs.  It did again this time, as the market obviously sold off hard once that slight complacency reading was reached.  After Friday's carnage, we're back up to 56%, so the complacency (if you want to call it that) is gone, although we're not even close to oversold.  The VXN has been stuck in a fairly tight range recently, so there's not much to say about that indicator.

NYSE MEMBERS REPORT - The most recent reporting period covered a time where the averages ended pretty much unchanged, so it shouldn't be too surprising to see the Specialist Short Ratio remain relatively unchanged at 38% (from 36% last week which apparently was correct).  The next report should be considerably more interesting, as the averages declined around 5% during the reporting period.  According to the correlation I pointed out several weeks ago, we can expect the SSR to drop significantly, possibly even reaching the 35% level that would be one full standard deviation from the 20-year median.  If that happens, I would consider it significant and quite bullish, as it would show that the public is once again jumping on the shorting bandwagon.

STEM.MR  - Since this model went into strong negative territory on Thursday, it's been working off that overbought extreme and is currently neutral.  Many of you emailed me on Friday asking why you hadn't received a positive indication since the market was down so much.  Well, it doesn't really matter how much the market moves, only our sentiment measures, and they just didn't get that extreme on Friday, regardless of market action.  That always tells me to not try to fight the trend and not expect any reversals, and indeed we never got any of significance intraday.  As we're not yet at an extreme here, I have no reason to expect an imminent upside reversal (considering this model only, and not our other factors).

I believe that last week's high and low will be incredibly important for the coming week.  If we break last Tuesday's lows, I expect a quick and severe selloff, and if we break Thursday's high, I expect some heavy institutional buying.  In the meantime, looking at the plurality of bullish indicators above, especially in the short-term, I would have to say that I'll be looking for the positive seasonality to exert some effect and create some positive momentum early in the week.  I would like to see a positive indication out of the STEM.MR model before considering short-term long positions, and if we get one we would likely be challenging last week's lows, so that could prove pivotal.  I believe we have enough positives from a sentiment perspective to create a sustainable rally at any time and I would consider a break of Thursday's high to be a good sign to get long, but we have not seen anything approaching historical extremes so a break of Tuesday's lows would have me on the sidelines (or short) until those extremes were reached.

 - Jason Goepfert