Extremes After an Historic Run

Bottom Line: We’re close to the average correction seen after extended bouts of a positive market. The real story is the lengthy, and quite often substantial, rally after such corrections work themselves out.

One of the main reasons I have been viewing oversold conditions to be so bullish is because I believe the trend persistency seen over the past year has to be respected. A couple of months ago I showed some statistics about the Dow Jones Industrial Average being more than 5% above its 200-day average for 191 consecutive days. This was an extremely rare feat historically, as it had happened only 15 times in the past 103 years. I took another in-depth look at these occurrences this weekend, this time looking at the first real corrections and subsequent performance after we saw such streaks. To get rid of some of the “double-counting” that may happen, I got rid of a few instances that were close in succession, leaving us with 12 distinct occurrences with which to work.

To show exactly what I’m talking about, below is a chart of the occurrence from 1943. Prior to this, the Dow had suffered a terrible six-year bear market, losing nearly half of its value. In the Spring of 1942, the market bottomed, and the Dow went into a very persistent uptrend (point 1 on the chart), ultimately logging 205 consecutive days at least 5% above its 200-day moving average (the red dotted line in the chart is 5% above the blue 200-day average). In July 1943, the market peaked and underwent an 11% correction over the next 95 trading days (point 2). Once the correction had run its course, the Dow then went on to a 16% gain over the next six months before finally undergoing another relatively tame correction (point 3).

The action in 1943 was fairly typical of how the market reacted after such displays of extraordinary price persistency. The table below outlines each of the instances since 1900, giving the number of days the Dow was able to remain at least 5% above its 200-day average, the number of days (and percent decline) of the first major correction after the streak, and the number of days (and percent rise) of the subsequent rally off the correction low before the next substantial decline set in.

Year	# Days in Streak	Correction		Subsequent Rally
Year	# Days in Streak	Days	%	Days	%
1905	220	26	15%	108	44%
1909	209	68	10%	188	26%
1922	238	31	10%	77	15%
1925	334	32	17%	97	23%
1929	194	16	10%	69	30%
1936	265	17	11%	217	35%
1943	205	95	11%	156	16%
1954	435	12	4%	124	18%
1959	300	38	9%	72	11%
1983	241	36	8%	81	12%
1986	219	17	10%	132	40%
1995	341	37	11%	166	38%
Average	267	39	11%	142	27%
Current	197	65	8%	?	?

From the table, we can see that the streak that ended this past March lasted 197 trading days, which is 70 days less than the average streak, though there were 5 other occurrences that lasted right around 200 days. So far, there have been 65 days in the correction that began in February, which is about a month longer than the average correction. The Dow has declined about 8% from the peak, which is a little less than the average correction of 11%.

If we assume that this instance will follow the average (big assumption, I know), then the downside risk appears to be relatively limited compared to the average gain of the subsequent rallies. There are two usual ways of measuring downside risk – either the average, or the worst drawdown. The average correction, as stated above, was 11%, and we are almost there. So that would mean that we could suffer another 3% or so before embarking on the next rally phase. The second method, which is much more conservative and uses the worst case, would mean that we would use the correction from 1925, which was a deep 17%. This would suggest that we could have another 9% to go on the downside. Both of these are very raw and are used simply as a rule of thumb, so there is obviously no reason why the Dow would have to decline more just to satisfy these rough assumptions.

Now turning to the reward side, we see that the average rally after the correction ran its course was a very healthy 27%, and it lasted around six months. In all cases, the rally lasted at least three months and gained at least 11%. There were five cases of rallies of at least 30%. Remember, this is AFTER the Dow had already spent at least 191 days at least 5% above its moving average. Also, I am being quite conservative in my reporting of the subsequent rallies. For example, in the 1943 instance, after the Dow began rising from the correction low, it didn’t undergo a 10% decline until 1946 (and a 64% gain). But to avoid “padding” the rally figures in the table above, I was very conservative in my measures of the subsequent rallies.

It is very clear from these precedents that the type of price persistence we saw last year does not go away easily. The market did not simply roll over into a new bear market – it usually suffered a mild correction (in terms of both time and price) – before embarking on another leg higher. When we combine the clear history of these examples with the gross oversold condition we reached on many of our measures last week, it reinforces the belief that the downside risk is limited compared to potential upside over the next few months.

Bottom Line: It’s true that recent breadth figures are skewed by rate-sensitive issues. But a look at the long-term advance/decline line shows us that rates have always had an effect.

Nearly everything I read or hear about the recent extremes in the advance/decline figures has been “asterisked” with the notation that interest-rate sensitive issues are dominating the numbers, so recent readings should be given less weight, or ignored altogether. While I too have suggested that recent readings are not exactly comparable to historical ones for various reasons, including the preponderance of rate-sensitive issues, I think it’s important to understand that rates have ALWAYS had an inordinate impact on the a/d line. This is not at all a new phenomenon.

The chart below shows the 30-year Treasury Yield versus a one-year moving average of the up issues ratio for the NYSE. The up issues ratio is what is posted to the site, and is simply the number of advancing issues divided by a total of advancing and declining issues – it is essentially the same thing as the advance/decline line.

There is a very clear correlation between peaks in yield and a bottoming of the a/d line, and bottoms in yield and peaks in the a/d line. While not perfect, this negative correlation between the two has existed for at least the past 27 years, and it is no surprise that the a/d line is peaking now as yields are rising.

So if yields are bottoming, and the one-year average of the a/d line is peaking, doesn’t that mean the indexes like the S&P 500 should decline? Not really. Ironically, peaks and troughs in the a/d line do not always coincide with peaks and troughs in the major market averages. The a/d line peaked in 1986, 1992, 1993, 1995 and 1997 while stocks continued to rally. Perhaps most egregiously, the a/d line formed a major low in 2000 – just as the major averages were putting in their peak.

Bottom Line: When volume in the S&P 500 ETF differs substantially from that of the underlying stocks, we often are about to see a trend change.

James Deporre of Realmoney.com posed an interesting question on Friday. The question was whether divergences between volume in an ETF and volume in the underlying securities lead to any kind of consistent market movements. It turns out that they may.

To check, I used total volume for all stocks in the S&P 500 and compared it to volume in the S&P 500 ETF (SPY, commonly known as “Spiders”). Unfortunately, I only have volume for the S&P component stocks going back to July 2002, but the data since then is interesting. As a check, I looked at SPY volume on any given day compared to its 50-day moving average. I then did the same for the component stocks of the S&P 500. Next, I compared whether SPY volume was relatively more or less heavy than the underlying stocks’ volume on a 10-day moving average basis. The following chart shows this difference.

When the blue line is above zero, it means that volume in SPY was greater than that in the S&P 500 component stocks, relative to their respective 50-day moving averages. For example, this past Wednesday volume in SPY was extremely heavy – by the close of trading, it was running 86% above what it had averaged over the past 50 days. On the other hand, volume in the component stocks of the S&P was running at only 17% above its 50-day average. This gives us “excess” SPY volume of 69%, which was the third-highest figure since July 2002. As we can see from the chart, when volume in the ETF becomes extremely high relative to volume in the underlying stocks, it has often coincided with market lows. Conversely, extremely low relative volume in SPY had an uncanny ability to show up near market peaks.

If we look at the widest spread between the two volume figures, we see that just this past March volume in SPY had far eclipsed that of the underlying stocks. This volume surge peaked on March 23^rd, the day before the low was put in. Lately, this has been happening again, as SPY volume is running about 20% higher than that of the underlyings’ (relative to their 50-day averages). While this could be considered somewhat high compared to all the other readings over the past two years or so, it is obviously nowhere near what we saw in March.

Block trading in SPY tends to run at about 30%-50% of total volume, which means that trades of 10,000 shares or more account for a good chunk of the volume. I don’t think there are very many individual investors who throw around those kinds of shares, so it’s safe to say the great majority of this is institutional volume. The ease of use and liquidity of these ETFs and the associated e-mini futures contracts are not lost on large accounts, as they can put on and removed hedges in a moment. I’m not going to bore you with my theories as to WHY volume flows into ETFs tend to be important indicators of market turning points, but I think it’s important to know that they are. Keep an eye on those volume figures – they could be an important clue going forward.

I think it’s safe to say that the technical condition of the market looks precarious at best. We’re seeing breakdowns from obvious support points, many indices are below popular moving averages, volume is usually higher on down days, etc. Most of those common technical reference points are suggesting we have more to go on the downside. But if we look objectively at what the market has done in the past after having seen a lengthy period of very positive performance (e.g. the Dow 200-day study above), it is clear that the market did not roll over into a new bear market right afterwards. Normally we saw a relatively mild correction, followed by another substantial leg higher. So far, we are seeing nothing that is significantly different from corrections past. Not only that, but we have seen historic levels of oversold conditions both in March and over the past week. All of those studies, and they are numerous, suggest that higher prices are much more likely than lower ones over the intermediate-term of 60-90 days plus.

One of the missing ingredients is excessive bearishness expressed by most of the sentiment surveys. These are certainly not perfect guides, as we saw last year, but it is a nice piece of the puzzle to have on your side. None of the surveys are showing excessive pessimism, and most of them are now neutral at best. We are getting some indications of excessive bearishness from some of the other sentiment measures, such as the put/call ratios which I have discussed ad nauseam. Assets in the short-sided Rydex funds have begun to generate quite a bit of interest, which is also marginally positive, and mutual fund outflows last week hit an extreme not seen since March 2003. Taken as a whole, I continue to believe that it makes sense from a risk/reward perspective to initiate or add equity exposure on weakness, with the intention of holding for several months.

In the short-term, our indicators are not giving much of a guide. The S&P appears trapped for the moment between 1090 and 1100. I imagine a violation of 1090 will most likely send us back to the lows while a break of 1100 would be the confirmation that many are looking for to gain confidence that we saw a tradable low last week. At the moment, all of our shortest-term indicators are neutral, not giving us a hint as to which level is most likely to break. Something that may influence which is more likely is the fact that we are going into option expiration.

As I’ve shown many times before, the week of option expiration has a tendency to be positive. Over the past 102 expirations, this week was positive 67% of the time with a minimal average return of 0.5%. I also checked to see if it made any difference if the week prior to opex week was positive or negative, and it did not – opex week still returned about the same percentage and was positive a roughly equal amount of the time. One additional factor may be the extremely high level of open interest in S&P 500 puts relative to calls. The SPX put/call open interest ratio is 1.75 (meaning there are 1.75 puts open for every call), which is the highest since 1997. If we look at how opex week performed when the open interest ratio was 1.50 or greater, the average return climbed to 0.9%, and 12 out of the 17 weeks were positive. Nothing to hang your hat on, but it may lend some support this week.

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.