Apr 13, 2003

Sunday, April 13, 2003

I've been suggesting that we would most likely see a trading range for nearly three weeks now. The S&P has traveled a total of 185 points since that time, but closed Friday with a total net change of a whopping 1.7 points. There were certainly trading opportunities during that time, but longer-term traders attempting breakout-type strategies most likely have losses to show for their efforts. We may remain mired in this environment for the time being, but from the looks of things the next sustained trend will most likely be to the downside.

Comment: Really nothing much has changed here from two weeks ago, so I suggest you check the March 30th weekly commentary for more background. In the large S&P futures contract, the positioning of the commercial traders vs. small speculators remains equity-friendly. Even with a market rise during the reporting week, commercial traders became more net long while small specs reduced theirs. This has brought all of the stochastics back to their most bullish positions possible and it continues to be a positive indication for the intermediate-term health of the market. Our one-year Futures Balance Matrix, which measures the aggression of commercial traders vs. small specs on the long side over the previous 52 weeks, has now been at 100% for four straight weeks. This is only the second time in the history of this data (back to 1986) where we have seen four weeks at 100%. The other time was for the period ending 4/29/92, however at that time this data was being collected only semi-monthly instead of weekly as it is now. The chart below shows all Matrix readings over 80% going back to 1986. The tan spikes that you see are readings over 80%.

You may have to squint a bit to see market reactions after such readings, but they have been quite positive. The average return three months after a reading of 90% or above is 5.0%, with the S&P being higher 86% of the time. This compares to a random three-month period with a return of 1.8% and 66% chance of being positive. Matrix readings of 10% or below, in contrast, give an average three-month return of -1.1%, with only a 49% chance of being positive.

Once again this week, however, we continue to see troubling positioning in the e-mini contract. You may know by now that I have found this contract somewhat effective only for the past six months, so I give it considerably less weight than the large contract. The most recent report shows small specs becoming their net longest in history, while commercials are close to their most net short in history. These stark differences appear to have a short-term dampening affect on the market (see the March 30th commentary).

Also troubling is the large positive variance in 30-year Treasury bond futures. Commercial traders are substantially more long than small speculators in this contract, and the difference grew larger this week. Since there is a very high negative correlation between equities and Treasury bonds (-.84 out of a possible -1.0 over the past three years), and a large positive variance in bonds very often leads to price gains in the bonds, this suggests that there is a high likelihood of declining equities. Again, you can reference the March 30th commentary in the archives for some more background on this.

Bottom Line: The conclusion here is the same as two weeks ago - short-term weakness is more likely than strength, but the intermediate-term outlook is quite positive.

Comment: We have only just begun to see our longer-term indicators turn around from their historically extreme negative readings. It will take considerably more time, probably another two weeks of a flat to declining market, before they enter neutral territory again. It's certainly possible for the market to rally after seeing such readings, but when in the midst of a longer-term downtrend, it's unlikely to see substantial progress when we've just experienced a rapid and extreme amount of public speculation. So, in the intermediate-term, these indicators remain bearish.

Shorter-term, we have seen enough assets shift out of the long funds and into the short-side funds that the RSI Spread indicator reached -91 on Friday. Even during this wicked bear market, readings of -91 or below have lead to a market with a five-day expectancy nearly 2% greater than that of a random period. What this means is that if you invest in the S&P 500 and hold for five days after we reach -90 (or below), you could reasonably expect to gain 2% more than you could by just picking a random 5-day holding period. What's most interesting is that strength immediately following readings under -90 tend to beget more strength, while imminent weakness portends even more weakness. Of the 5-day periods that were positive, the day immediately following the -90 day was also up 80% of the time. But of the 5-day periods that were negative, the 1st day after the -90 reading was up only 29% of the time. Put another way (perhaps more clearly), if Monday closes up on the day, then there is an 80% chance that we will still be higher by next Friday. However, if Monday closes lower, then there is only a 29% chance that we will be higher on Friday. Also, the bigger the move on Monday, the more positive or negative we should be by Friday. The sample size is relatively small here, so I certainly wouldn't trade based off this alone, but I think it provides something of a useful framework for approaching the week.

Bottom Line: If we see strength on Monday, I wouldn't necessarily be quick to fight it, as readings of -90 or below in the RSI Spread indicator have tended to lead to strong markets when the day immediately after is positive. If we see weakness on Monday, then additional rally attempts later in the week are more likely to fail than not. Longer-term, we still need to see a lot of speculation wear off of these funds before they could even be considered neutral.

Comment: The VIX and VXN, particularly the VIX, are at interesting junctures. As we can see from the VIX Fear Premium posted to the site, there was a gargantuan amount of uncertainty being priced in prior to the war with Iraq. it will take time for that to unwind, and we still have a long ways to go before the VIX could be considered to be showing complacency in terms of the market's historical volatility. However, I also like to monitor the VIX in terms of its relation to its 10-day moving average, and in that case it can be considered mildly overbought. A slowly declining VIX actually tends to be bullish for the market, but when it spikes lower and becomes stretched from its average, then that's where trouble comes in. So while its current level should be watched closely, I would not consider it extremely negative for the market until it begins to make a reversal higher.

That said, I want to point out another aspect of the VIX which is troubling. I should warn you that this may get a little complex, but please stay with it because I think it's important. Volatility is mean-reverting, meaning that periods of high volatility tend to follow periods of low volatility. There are a ton of ways to monitor these cycles, and the VIX is one of the better ones. For a longer-term perspective of how much complacency the VIX may be pricing in, we can look at how far the 10-day average of the VIX is stretched from the 200-day average. Currently, the 10-day average is about 13% below the 200-day average, and that's quite low (bearish for the market). However, if we want to see just how extreme it is on a standard deviation basis, we are presented with a serious problem, and it's a common one for sentiment measurements. The problem is that many sentiment indicators, the VIX in particular, have a very high statistical skew. You don't need to have an understanding of statistics to see that the VIX often has very high spikes higher, but very few spikes lower. It regularly spikes above 40 or even 50, but very rarely goes below 20 (particularly recently). The following table presents the total number of occurrences of the VIX exceeding one or two standard deviations on a closing basis since 1986:

DEVIATIONS	OCCURRENCES
+ 1 S.D.	521
- 1 S.D.	548
+ 2 S.D.	152
- 2 S.D.	0

We can see that it exceeded its upper standard deviation band (i.e. +1) a total of 521 times. It exceeded its lower band 548 times. This is relatively close, and should be expected if we have a normal distribution. However, and this is where the problem comes in, the VIX exceeded its 2nd upper deviation band 152 times, but never pierced its 2nd lower deviation band. With this great of a skew, it basically renders the idea of standard deviations useless. Fortunately, there is a way to correct for this phenomenon, which is where the chart below comes in. The statistics behind the chart are not important, but the concept is - when the VIX gets to the +10% level, it is oversold (and the market overbought) in a statistically meaningful way. Conversely, a reading below the -10% level suggests the VIX is overbought (and the market oversold). This allows us to determine the degree of overbought/oversold without the distorting effects of the traditional skew of the indicator.

We can see that when the indicator reached its upper extremes, the market had a difficult time making headway, even during the latter part of the bull market. When it reached the lower extremes, the market environment tended to be positive for equities, even during the bear market. Currently, the indicator is once again more than +10%, for the first time since January. We have to go back to May 2002 for the next previous occurrence, and that of course was also not exactly healthy for the market. Looking at the VIX in this manner suggests that equities are overbought in a historically meaningful way, and further rallies should be looked upon as selling opportunities.

Bottom Line: The traditional way of looking at the VIX is troublesome enough (especially if it begins rising again), but if we transform it into a way where it can be compared apples-to-apples with historical extremes, it becomes even more so. While the VIX is somewhat inflated in relation to actual volatility, it is extremely low in an historical, mean-reverting context.

Comment: The 10-day up issues ratio and up volume ratio spent all week bouncing around the upper (overbought) end of neutral. This week, we'll be dropping two large positive readings on Tuesday and Wednesday, so it's possible we could head closer to oversold should we have an overall down week. However, the 21-day moving averages, which I have discussed several times over the past week, are still ominously overbought and that will take more time to work off. This has historically been a very effective contra-trend indicator (meaning it does a good job at identifying overbought conditions in a downtrend, and oversold situations in an uptrend), and since we're still in a downtrend, it argues strongly that lower prices are much more likely than significantly higher ones.

Our Down Pressure indicators have worked quite well during this trading range environment, as have most oscillator-type indicators. They became oversold on 3/31 then switched to overbought on 4/3. Since the indicators posted to the site are a 3-day average of the daily readings, and we haven't put together even a 3-day trend for awhile (or seen any extreme individual readings), the indicators for both the S&P 500 and Nasdaq 100 are currently neutral. You may have noticed that they are on the upper (oversold) end of neutral, and look to become oversold, but tomorrow we'll be dropping the fairly large reading we saw from last Wednesday, so it would take an equally bad day just to keep us where we are. Only if we have large down markets on Monday and Tuesday will these indicators become oversold.

All of the cumulative TICK indicators posted to the site are currently neutral. However, I also keep track of several longer-term intraday indicators which I have not posted. The one I have attached below shows movements during the bear market and uses intraday data (30-minute), but is a 21-day moving average. So, it observes how the TICK behaves intraday over the past trading month. I have circled each instance of the indicator peaking over the +60,000 mark.

Obviously, such bouts of extended buying pressure have not been sustainable, as weakness has followed each instance. Let's take a look at this another way - cumulative TICK as a function of price:

This simply takes the cumulative TICK and divides it by the current value of the S&P 500. We can see the steady uptrend since the bear market began. In fact, we just hit a new high on Friday. Even though price has continued to erode, there has been no letup in the buying pressure once price appears to recover.

Let's explore that aspect just a bit further. The following chart looks at the number of days that see a +1000 TICK reading versus those that show a -1000 TICK over the previous 30 days. Each day that shows a +1000 reading gets a score of +1, while those that show -1000 get -1.

We can see that the low in October 1999 saw a score of -9, meaning that over the previous 30 days, there were 9 more days that showed a -1000 TICK reading than showed a +1000 TICK. Currently, we just hit a new high, as the previous 30 days have shown 16 more +1000 days than -1000 days. During this entire time frame, the total number of issues traded on the NYSE has remained fairly steady, so that does not bias the data at all. As we progress through the bear market, we will become more and more prone to buying panics, as traders believe that "this time has GOT to be different" and panic to get long, hoping to catch the bottom of the bear market. Even when the increasing amount of program trading (the automatic buy and sell programs issued by institutions) is factored in, it does not distort the above charts by a significant amount.

The 10-day TRIN readings for both the NYSE and Nasdaq are relatively neutral. If we have an up day tomorrow, they have a decent chance at making a dive towards overbought, but we will likely not see any extremes in either indicator this coming week.

Bottom Line: The longer-term breadth of the market is seriously overbought. The 21-day averages of the a/d line, up volume and cumulative TICKS are all at areas that have identified grossly overbought conditions historically. While we could certainly have small rallies from this point, a sustained upward trend is highly unlikely.

Comment: While still historically low, the Specialist Short Ratio rose quite dramatically for the week ending 3/21/03 and didn't drop much the next week even though the S&P 500 declined by nearly 4%. This ratio shows the percentage of NYSE short sales made by smart-money specialists on that exchange compared to the public, so a low ratio means that specialists are buyers of stock and the public is shorting heavily - this is bullish. As short-oriented hedge funds proliferate, short strategies become more popular (and convenient) for retail traders, and the bear market continues, this ratio has shown a steady decline. This secular decline has been in place really for the past 40 years, but it has especially taken hold over the past 2 years. This isn't an especially troublesome problem as far as market timing goes, as long as you find a way to take that trend out of the ratio. One of the ways I like to do this is by looking at the rate of change of the ratio. Rate of change simply looks at the current reading compared to the reading "x" periods ago. For this data, I find that an 8-week R.O.C. makes sense, so let's take a look at that:

We can see from the chart that in late January and again in early March, the momentum was towards a lower ratio. The rate of change was negative, and close to one standard deviation from its all-time historical mean (the thick green horizontal line on the chart). This just means that those readings were about 15% below the readings 8 weeks prior. Generally, the risk/reward at that time was skewed to the long side, as the wrong-way public was becoming quite aggressive in their shorting activities. You can also see, however, that the pre-war rally leading up to March 21st was greeted enthusiastically by the public, as they greatly reduced their shorting activity relative to the specialists. In fact, total public shorts increased about 20% that week, while specialist shorts increased nearly 60%. This caused the rate of change to jump to +18% - a sign of momentum that has marked intermediate-term peaks in equities, especially during this bear market. As of March 28th (the most recent data available), the rate of change settled back down to 0%, which is exactly neutral. Next week, we should see the R.O.C. spike over +15% once again, as we rallied during the reporting week (which most often causes the Specialist Short Ratio to rise), and the reading 8 weeks prior will have been quite low.

Bottom Line: The recent developments in the NYSE Members Report show that while public shorting is still relatively heavy, the momentum has waned considerably. In fact, the momentum has dropped so much that it could be considered bearish. Next week's information should shed some additional light on the situation, as we will be able to see how the public reacted to the rally after the first decline of the early March jump in equities.

Comment: Most of the surveys we monitor have returned to neutral territory after the slight negative extremes they reached in late February or early March. This is not particularly bearish as far as I'm concerned, as it's what we should expect after a 10% move off of a low. I did a comparison two weeks ago which showed how the readings have changed during this move as opposed to other intermediate-term lows. The current readings were on the high side, and overall they've increased only a small amount since then. I've seen multiple references suggesting that since the Investor's Intelligence survey is showing the percentage of bulls greater than 50%, it's a negative for the market. I assume the rationale for such statements is that every time it has been above 50% since 2000, the market has shown weakness. However, most of the instances in which the bulls have been above 50%, the bears have also been BELOW 30%, often by a substantial amount. Currently, the bearish percentage is over 31%, and would need to drop several more percent before it would seriously trouble me. Situations like this are why I prefer to monitor measures which take into account both the bullish and bearish percentages together (e.g. the "bull ratio" which is posted to the site). Currently, the bull ratio for the II survey is just over 62%. Most of the prior peaks saw readings well above 65%.

Bottom Line: The level of bullish opinion we're seeing expressed in the sentiment surveys is certainly not positive for the market. However, it's close to what should be expected after a 10%+ rally from a low, and is not particularly bearish either. Until investor opinions become extreme once again, or do something they "shouldn't" do, there's not much to glean from this information.

Comment: The unwinding of the negative momentum in the sentiment surveys since the low in March has served to bring this model back down to the 50% level, from the extreme of 66.5% it reached in early March. Notably, the model is beginning to rapidly approach its lower trading band, which is based on mean values of the model over the past two years. Currently, there is only about 4% separating the model from its lower band, which is the least amount of "wiggle room" we've seen since March and November/December last year. This shouldn't yet be considered extreme, or even particularly bearish, but it should be monitored very closely because another two weeks - or even one - of increased bullishness in the surveys could bump this model into negative territory.

Bottom Line: While this model is not yet to a point that would have me on edge, it's getting close. If we see even mediocre jumps in bullishness over the coming week(s), the model will likely dip under its lower trading band, which has been a death knell for equities over the past few years.

Comment: Everyone's favorite day, April 15th, will be kind enough to visit us this week. Since 04/15 was made the official day to pay our dues as citizens in 1955 (it was March 15th prior to that), the market has shown just a pinch of a bias. The S&P 500 return the day taxes are due (the 15th if it falls during the week, and the following Monday if it falls on a weekend) has averaged 0.4% over the past 48 years, with 67% of those days being positive. This is substantially better than a random day. Interestingly, we have seen 7 "tax due" days close higher by more than 1%, but not one close down by more than 1%. The days immediately surrounding "tax due" days showed a very slight positive bias (each of them have been up about 60% of the time), but probably not enough of one to be considered significant. There have been six years where the 15th fell on a Tuesday, as it does this year, and the day before (Monday) has been up an average of 0.5%, with 4 of the 6 days closing higher.

We also have Good Friday this week, so next Thursday will put an end to this coming week. Easter is not a particularly strong holiday market-wise, as the table below illustrates:

	UP	DOWN	UNCH	SUCCESS RATE	AVG. RETURN
DAY BEFORE	33	11	9	75%	0.23%
DAY AFTER	17	26	10	39%	-0.25%

We can see that the trading day before the Easter holiday has been up 33 out of the past 53 years (unchanged is considered anything less than a 0.1% return, positive or negative), for a "success rate", or percentage of time positive (ignoring unchanged years), of 75%. The trading day AFTER Easter, however, does not often fare well, as the average return is negative and the success rate is only 39%.

There has been a fairly strong trend to this holiday recently. Here is how the last 10 years have fared around Easter. For this table, all returns (even those that would be considered "unchanged" above) are considered.

YEAR	DAY BEFORE	DAY AFTER
2002	UP	DOWN
2001	UP	DOWN
2000	UP	DOWN
1999	UP	UP
1998	UP	DOWN
1997	DOWN	DOWN
1996	DOWN	DOWN
1995	UP	DOWN
1994	UP	DOWN
1993	DOWN	UP

For demonstration purposes, if you had bought the S&P 500 at the close the day before the last trading day before Easter, sold on the close the next day, then sold short at that same close and covered the day following Easter, you would have made $508 from a $10,000 investment. This is a return of more than 5% while being in the market only 20 of over 2,500 trading days. I would not recommend anyone try this, but I think it's useful to keep this bias in mind, especially for very short-term traders.

We have option expiration this week as well. Expiration weeks have tended to have a slight positive bias early in the week, then weakness later (particularly expiration Friday, though the holiday this week renders that pointless). A few weeks ago, I discussed the results of a study looking at volatility during these weeks. Contrary to popular opinion, they are no more volatile than any other week (as far as volume and daily range goes), so don't buy into the media hype blaming expiration for whatever move happens to occur.

Bottom Line: It's tough to pin too much hope on marginal biases, so I would still say overall that the seasonality this week is neutral. If pressed, I would suggest that we may be a bit more inclined to see some strength early in the week, what with expiration and tax day apparently having some sort of positive effect on the market, though that is very tenuous. With a bit more gusto, I would suggest that we may see some strength on Thursday, with weakness to follow the next Monday.

Comment: I consider the OEX put/call ratio to be a non-contrarian indicator, meaning that a high volume of calls relative to puts could be considered bullish (the exact opposite of the equity p/c ratio). Currently, the 10-day average of the OEX ratio is quite low and that would normally be a bullish indication for the market. However, over the past six months or so, this indicator has been ineffective, so the weight I give it is minimal at best.

The other put/call ratios I monitor are all neutral at the moment and not giving us much direction.

Bottom Line: The most effective ratios here are neutral. We are getting a positive reading from the OEX ratios, although that has not been effective recently, so the weight it is given should be diminished.

Comment: The utter lack of a sustained trend has prevented our shortest-term model from reaching any kind of extreme since April 2nd. There is just not much to say here.

Bottom Line: Until we get some kind of trend, this model will likely stay sidetracked.

Comment: This model has remained neutral for quite some time and will likely continue to be so this coming week as well. Notably, the lower (bearish) standard deviation band has been curling up, and is 10% higher than it was a year ago. This means that it will be considerably easier for the model to become overbought that it would have been at this time last year. However, until we see more optimism from the sentiment surveys, and some negative changes in the Commitments of Traders information, it will be a struggle to get the model to overbought.

Bottom Line: Still neutral here, and will likely continue to be so for the foreseeable future.

Comment: As is stated on the site, I tend to interpret this model in terms of standard deviations from its recent mean. However, if we step back a bit and look at its absolute level compared to those over the past year, the outlook is grim.

Each time the model dropped under the 30% level is circled. Currently, we are well below that level, and as we can see from the chart, the precedent is not encouraging for the long side.

Bottom Line: Two weeks ago, l said that the model's current positioning suggested that a meaningful rise in equities should be met fairly quickly with an overbought reading in the model. Conversely, a market decline would not be met with oversold readings for quite some time. This suggested that the market had limited prospects on the upside compared to downside, and that remains the case today. According to this model, any upside should be used as a selling opportunity.

This week, we have few bullish indications. About the only intermediate-term positive for the market is the action in the large S&P futures contract. In the short-term, positives include the historically decent action following quick shifts into the bearish Rydex funds (which has caused the RSI Spread to spike down to -91), and the positive seasonality that should be evident on Thursday (and possibly tax day Tuesday). Other than those few positives, everything else we monitor appears neutral at best and exceedingly bearish at worst. This suggests to me that while we could see some upside attempts this week, they will most likely be met with selling pressure. Therefore, I would use higher prices as an opportunity to sell or sell short equities. Each time I base such an opinion on historical precedents, I get several emails asking if this time could possibly be different. As always...yes! There is *always* the possibility that this time will be different than most others. We could be beginning a new bull market. But I have made much more money by believing things will remain the same than I have believing this time isn't like all the others for some reason. I like to keep things as simple as possible, so until the major downward trendline on the S&P weekly chart is broken, I believe the bear market is intact and rallies should be sold.

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.