Barron's Major Market Top Update
In the May 4, 1998 edition of Barron's, Peter Eliades wrote an article describing his "Sign of the Bear" indicator which gave a signal in early April 1998. Within a few weeks of the signal, The Value Line Composite Geometric Index, arguably one of the most accurate indicators of the performance of an average investor's portfolio, reached a new all-time high. To this date (March 2004), that high has not been surpassed on the Value Line Composite Geometric Index. Since the "Sign of the Bear" signal in 1998, another signal was given in September 2000 within days of the all-time high in the NY Composite Index. That high also remains the all-time high to this date. The market action after September 2000 in the Nasdaq Composite and the S&P 500 makes it clear that the September 2000 signal was another fabulous success for the "Sign of the Bear."
The two articles below were prepared in 1998. The first link is the unedited article written on our development of the "Sign of the Bear" indicator.The second link is the revised indicator as described in the Barron's May 4, 1998 edition.
The Sign of the Bear, April 1998 (edited)
April 28, 1998
Addendum to SPECIAL report, "The Sign of the Bear"
Since writing "The Sign of the Bear" in mid-April, we have made some new discoveries relating to the churning pattern discussed therein. Barron’s magazine has shown an interest in our research and we have been led to believe they will run a story based on the pattern in the very near future, perhaps as early as the issue dated May 4. We are sending you the updated story and research. It takes nothing away from the original research, but it narrows the parameters for the "The Sign of the Bear." The ones that remain are truly five of the most important tops of this century. We hope you enjoy our updated version.
Barron’s Sign of the Bear article
It is the seemingly impossible dream of market technicians to discover an indicator or pattern that would consistently point to major turning points in the stock market, and although it is doubtful such an indicator exists, we have discovered a pattern that has been remarkable in its ability to discern approaching market tops. We stumbled upon the pattern in November 1992 because we were struck by the apparent lack of volatility in the daily number of advancing and declining issues on the NY Exchange. Over a period of 21 trading days (the number of trading days in the average month), the highest single day advance/decline ratio (simply divide up stocks by down stocks on the NY Exchange after the close) was 1.84 and the lowest was .71. Rather than use those exact limits as a precedent for further research, we arbitrarily widened the limits somewhat to .65 and 1.95. We first searched our computer database for other time periods of 21 trading days when similar "churning" occurred, periods that were noteworthy for their lack of volatility. Initially we went back to 1966 when the Dow made its first move to the 1,000 level. We were stunned by the results.
From January 1966 to November 1992, a period of almost 27 years, there were only three other periods when the above conditions were satisfied, namely, over a 21 day period there was no daily ratio higher than 1.95 and no daily ratio lower than .65. Here are the only dates when those conditions were satisfied:
January 25, 26, 27, 28 1966
October 17, 18, 21, 22, 24, 25 1968
December 6, 7, 8, 11 1972
If you are acquainted with market history, you will recognize those immediately as time period just weeks before three of the most important market tops of the past 30-40 years. Just how important were they?
The final 1966 top occurred on February 9, eleven trading days after January 25,the 21st day of the churning pattern. On an inflation-weighted Dow chart, the 1966 top stood until 1995 as an all time Dow high.
In 1968, the pattern lasted 26 days. The final Dow high came on December 3, 27 market days after the churning pattern became official on October 17, its 21st day. From that 1968 high, the average share of stock and equity mutual fund lost 70-80% of its value over the next six years.
The next time period to see 21 consecutive churning days was December 6, 1972. The final high in that time period occurred on January 11, 1973, 23 trading days after December 6. That top led to one of the sharpest two year Dow declines in history, almost 50% in less than 24 months.
After reviewing those results, we felt we had discovered something special. During a period of almost 27 years, there were only three occurrences of the pattern and each occurrence led to a major market top within, at most, 27 trading days. We ultimately dubbed the pattern "The Sign of the Bear."
We have learned in almost three decades of market research that any pattern that appears to have predictive potential should be researched as far back as is practicable. Without going into the detail of the sequence of our research from that point, we found that the period from 1940 to 1966 contained nine churning patterns. Most of them led to bearish resolutions, but none of those predicted tops ranked with the importance of the three tops detailed above, and some of the patterns in that time period did not lead to adverse market resolutions. When we worked backwards from 1940 to the beginning of our daily advance/decline data in 1926, however, we were once again stunned by the results. The first 21 day churning pattern prior to 1940 occurred on July 19, 1929. That’s right! Just over six weeks prior to the most famous top of this century, the "Sign of the bear" appeared.
Our final research brought the data up to date through the current time period. In all, from 1926 to early April 1998, a period of over 72 years, there were only 19 churning patters of 21 days or longer. There were two long stretches of time, from mid 1929 to late 1952, and again from early 1966 to late 1992 when there was no one churning pattern. We bring this up to show the relative rarity of the pattern. Our problem in analyzing the nineteen prior churning patterns was to find potentially distinguishing characteristics between those churning patterns which led to important tops and major declines and the patterns which led to only short term declines, or worse yet, no decline at all. We have indeed discovered two such characteristics that have accompanied all the churning patterns that led to major market tops. The first characteristic is that the pattern must continue for no longer than 27 trading days. At first, it seemed to us intuitively that the longer the pattern ran, the more negative the ultimate repercussions should be. It just wasn't true.
Five of the 19 patterns were longer than 27 trading days and none of them led to important market tops. Three of those five have occurred in the last six years. The second characteristic was the history of the pattern just after the consecutive days of churning ended. The following three days determine the predictive importance of the pattern. If the pattern is to lead to a major top, the two or three days after the churning pattern ends should be quite negative and we can quantify how negative they must be. Calculate the daily A/D ratio (advances divided by declines for the three days following the end of the pattern. Use both an average of the first two days and an average of the first three days after the pattern ends. If either of those averages is less than .75, the pattern becomes a "Sign of the Bear." Those two requirements eliminate 14 of the 19 prior churning patterns. Here is what is left:
Date Days of churning Lowest Ratio (2 or 3 Day) Subsequent Decline
July 19, 1929 22 .733 89.5%
December 8, 1961 24 .647 29.1
January 25, 1966 24 .396 26.5
October 17, 1968 26 .582 36.9
December 6, 1972 24 .585 46.5
April 6, 1998 21 .592 46.5
As you can see from the above table, a 21 day churning pattern ended on April 6, 1998. It fit both additional qualifications given above, making it only the sixth "Sign of the Bear" in the past 72 years. Be aware that the "Sign of the Bear" usually precedes the final market high; it does not coincide with or follow the market top. When we presented these data to subscribers the week of April 6-10, we noted that if this pattern resolved in a similar fashion to the last three resolutions the Dow would see a final high close of 9182.90 +_ 60 points. As technicians, all we deal with is probabilities. Be aware that although the "Sign of the Bear" has been rare in the past and effective in predicting market tops, it has been discovered after the fact, and we have never used it to predict a top in real time. It should not necessarily prompt you to sell all your stocks tomorrow morning. It should, however, caution you against the "new paradigm" fundamentalist outlook that nothing can go wrong with the market for at least several years. We can assure you that there were no fundamental indications of a market top at any of the above prior signals. Here is a quote from one of the prior time periods"
"This prosperity might be disquieting if it were accompanied by any of the symptoms of inflation such as swelling inventories or a growing urge to buy for more than immediate requirements. It is also encouraging to see that despite the continuance of prosperity, commodity prices remain stable." –Outlook and Independent 8-7-29
Because our cycles allow for a final top as late as August 1998, we have to allow for that possibility. If the market is making new highs beyond the end of May, then the power of the April signal will have dissipated. In that case we would look for the possibility of another "Sign of the Bear" prior to a July or early August top. But be on guard. The "Sign of the Bear" has been seen in this time period. It has not been wise to ignore it in the past.
The Sign of the Bear, April 1998 (unedited)
The seemingly impossible dream of market technicians is to discover an indicator or pattern that would consistently point to major turning points in the stock market. It is doubtful such an indicator exists, but we have stumbled upon a pattern that qualifies as remarkable in its ability to discern approaching market tops. Because of the market action that often follows this pattern, we have called it "The Sign of the Bear."
In November 1992, we were struck by the apparent lack of volatility in the daily advance decline statistics on the New York Stock Exchange. We examined the daily advance/decline ratio (simply divide advances by declines on the New York Stock Exchange e.g. 1800 advances, 900 declines = A/D ratio of 2:1) and discovered that over the prior 21 market days, the highest ratio was 1.84 to 1 and the lowest ratio .71 to 1. We used 21 market days because there are that many trading days in the average month.
Rather than use those exact limits as a precedent for further research, we arbitrarily widened the limits somewhat to .65 and 1.95. We then searched our computer database for other time periods of 21 market days when similar "churning" occurred in the market, periods that were noteworthy for their lack of volatility. We started by going back to 1966 when the Dow made its initial move to the 1000 level. We were stunned by the results.
From 1966 to November 1992, a period of almost 27 years, there were only three other periods when the conditions were satisfied, namely the highest daily A/D ratio over a 21 market day period was less than 1.95, the lowest ratio over the same period was greater than .65. Here are the dates:
January 25, 26, 27, 28 - 1966
October 17, 18, 21, 22, 24, 25 - 1968
December 6, 7, 8, 11 - 1972
Note that the first date given in each sequence marked the 21st consecutive day of churning as defined above (daily A/D ratio smaller than 1.95 but greater than .65). From that, we can see that in 1966, there were 24 consecutive days before the streak was broken, in 1968, 26 consecutive days, and in 1972, 24 consecutive days. Anyone even marginally acquainted with stock market history should recognize the above dates immediately. In each case the dates preceded very important market tops by no more than five weeks.
The final 1966 top occurred on February 9, 11 market days after January 25, although the final top day closed only 4.51 points higher than the first day to meet the 21 day churning specifications, namely January 25. On an inflation-weighted Dow chart, February 9, 1966 stood until 1995 as an all-time Dow high.
The first day of the range of days from October 17-25, 1968 preceded the top of that market by 27 market days and 2.7% on the Dow. From the 1968 closing high on December 3, the Dow collapsed almost 36% over the next 18 months. Perhaps more importantly, the average share of stock as measured by the Value Line Composite Index lost over 75% of its value between December 3, 1968 and October 4, 1974.
The next time period to see 21 consecutive churning days occurred on December 6, 1972. The final high in that time period occurred on January 11, 1973, 23 market days after December 6 and 2.4% higher on a closing basis. That top led to one of the sharpest two year market declines in history with the Dow down almost 50% in less than 24 months.
You can understand that, after reviewing these results, we thought we had discovered something special. From 1966 to November 1992, there had been only three time periods when the churning parameters were met for 21 consecutive trading days or longer and each of them was a precursor to three of the most important market tops of this century. Through almost three decades of market research, however, we learned that any indicator or pattern should be researched as far back as practicable, so our subsequent research took the search for similar churning patterns back to 1940. With the exception of the period from June 1963 to March 1965, the results were impressive, though not as uniformly dramatic as the post 1965 results noted above. Here are all the similar periods noted from 1940 through 1965 where the 21 day churning pattern occurred, followed by the number of days in the pattern.
December 30, 31 ‘52, January 1 1953 (23)
September 22, 23 1955 (22)
May 3, 6, 7, 8, 9, 10, 13 1957 (27)
December 8, 11, 12, 13 1961 (24)
June 7 1963 (21)
February 28 through April 23 1964 (59)
February 4,5 1965 (22)
March 16-25 1965 (28)
May 11, 12, 13, 14 1965 (24)
For now, let's discard the period between June 1963 and February 1965 and observe the average results for the remaining six periods. The Dow, on average, advanced 1.7% from the close of the 21st day in the pattern to the highest subsequent intra-day high after the pattern emerged. It took ten market days to reach that high on average, and the subsequent decline averaged 15.2%. We deem those results significant and impressive.
A market historian would note a remarkable commonality in all of the above periods. Almost without exception, each time the churning parameters were satisfied over a minimum of 21 trading days, the market was either at an all-time high or a multi-year high. There is nothing in the definition of the two parameters required over a one month period that would suggest such a result.
Before we get to the recent history of the pattern, let’s review the data prior to 1940, data that we only gained access to recently, so they were not included in the original study of the pattern. Remember, after our initial discovery of the pattern, our follow-up research went back to 1940. There was not one instance of the pattern from 1940 until December, 1952. But when we received the data prior to 1940, we were again stunned by the results. There was not one instance of the pattern in the decade of the 1930s, just as there had been none in the 1940s. Working backwards from December 1952, take a wild guess when the first appearance of the pattern occurred. The dates were July 19 and 20, 1929. That’s right! Just over six weeks before the most famous top of this century, the "Sign of the Bear" appeared. It would not be seen again until 1952, over 23 years later. There were 6,867 market days between those two incidents of the pattern.
That leads us to the history since 1968. Remember that the 1929 top, arguably the most important top of the century, was closely preceded (46 calendar days) by the pattern, and remember also that the pattern did not reappear for over two decades after the 1929 top. The 1968 top led to a silent crash, due both to inflation which did not show up directly in the market averages, and due to the fact that the average share of stock declined almost 80% from 1968 until 1974. Should it surprise us that the churning pattern also disappeared between 1972 and 1992, again a period of around two decades? Just as in the 60s, the 1990s have now produced a series of churning patterns, most of which have had no apparent effect on the market. Let us give you the dates for all of the 21 day patterns since 1972. The numbers in parentheses after the dates represent the number of days that the churning pattern persisted.
November 9 to December 17 1992 (48)
August 20 to September 3 1993 (31)
January 18 to February 3 1994 (33)
April 25 to May 2 1995 (26)
September 12 to September 15 1995 (24)
Those have been the only five churning patterns of 21 days or longer in the past 26 years, prior to April 1998. The 1994 pattern accompanied the January 1994 top, but the other four were non-events, much like the patterns of 1963 to early 1965. In looking back at all the patterns and attempting to distinguish the characteristics of patterns which closely preceded significant market declines, one of the items that appeared significant was the length of the pattern. Intuition would suggest that the longer the pattern, the greater its potential negative influence. History has proved otherwise. Once a pattern moves beyond 27-28 market days, it has far less a chance of being significant. The 1992 pattern can be eliminated because it was far too long and, more importantly, it did not fit the profile of prior patterns which saw the Dow going to either multi-year highs or all-time highs as the pattern reached 21 days. It emerged at a time when the Dow was more than 5 months beyond and almost 6% lower than its previous all-time or multi-year high. It just did not fit into the profile of prior churning patterns. The April-May 1995 pattern and the September 1995 pattern appeared at first glance to qualify as patterns that led to intermediate or long term tops in the past. As we further researched successful predictive patterns of the past, however, we noticed a signature that accompanied almost all the successful pattern predictions of market tops. One important consideration was how the pattern ended. In other words, when the churning streak ended, did it end with a high ratio (above 1.95) or a low ratio (below .65). The four very major tops preceded by the "Sign of the Bear," namely 1929, 1966, 1968, and 1972-73 all ended with low ratios. In fact, they not only ended with low ratios, in three of the four instances, there was a low ratio for two consecutive days to end the pattern. In the fourth instance, there was a low ratio in two of the next three days. Since the 1972-73 pattern, the five patterns listed above have resolved as follows:
November 9 to December 17, 1992---Plus day resolution
August 20 to September 3, 1993--Minus day resolution-2 days
January 18 to February 3, 1994--Minus day resolution-1
April 25 to May 2, 1995--Plus day resolution
September 12 to September 15, 1995--Minus day resolution-1
It is impossible to say how important the positive or negative resolution is relative to predicting a potential top of importance. The number of occurrences is not statistically significant, but from the data we have, it is clear that the most important tops in the past have all been preceded by minus day resolutions (advance/decline ratio smaller than .65) that occurred in at least two of the three days after the churning streak ends.
As this report is being written, a new streak has just been formed and broken. The streak was ended on April 6, 1998 after it reached 21 consecutive days for only the 6th time in the past 26 years. It occurred in a time period when we expect the potential resolution of an important market topping cycle (between April 28 and August 22, 1998). Importantly, the streak was broken with a down day (A/D ratio=below .65, specifically .415).
Remember, the ending of the pattern usually precedes the final market high; it does not usually coincide with or follow the market top. That would mean that after the ending of the pattern, we should typically expect higher highs. If we take a profile of the three most recent important tops (1966, 1968, and 1973) preceded by the churning pattern, the average resolution would occur as follows. The pattern would end with at least two out of three days having an advance-decline ratio lower than .65, or a two day average ratio, following the last day of the streak, below .60. Because the last churn day is now known to be April 6, if the pattern resolves in an average fashion, there will be a decline to around 8785 intra-day, ending around April 14-15, then an 11 day (trading days) market rally to April 28-29 to a potentially important market top. The top should be around 4.8% higher than the low close between the April 6 close and the subsequent top due in late April or in May. Another way of calculating a potential top would be to base it on the average final advance from the close of the last day in the pattern. In this case, the average resolution would call for a final Dow close of 9182.9 plus or minus 65 points.
By the time many of you receive this special report, the resolution will already have occurred. Don’t expect it to occur (or have occurred) exactly as outlined above. The important fact to note is that "The Sign of the Bear" has again been seen. It is a rare pattern that has preceded most of the important market tops of this century.
The following pages contain charts of almost all the incidents of "The Sign of the Bear" since 1929. We have eliminated the patterns of June 1963, February-April 1964, November-December 1992, August-September 1993, January-February 1994, and both of the patterns in 1995. Of those, the only patterns arbitrarily eliminated were June 1963 and September 1995. The 1964, 1992, 1993, and 1994 patterns were eliminated because of their length. As we noted earlier, longer patterns were not, as might be presumed, more bearish. In fact, the most effective patterns in predicting a bearish resolution were almost all shorter than 28 days. The one minor exception was the 1994 pattern which was 33 days long, but still accompanied a market top that lasted for over a year and led to a 10% decline. The April-May 1995 pattern ended with a positive breakout (daily A/D ratio greater than 1.95) and we noted earlier that all prior successful predictors of bearish resolutions ended with a negative resolution (A/D ratio smaller than .65).
We will begin with the "Big Four," the patterns that preceded four of the most important tops of this century.
The final chart is the current pattern which ended on April 6, 1998, after 21 days of churning. Its personality appears to be similar to that of those which either closely preceded major market tops over the past 70 years or led to important intermediate term declines. In association with the long term cycle that is due to produce an important market top within the next 4 months, it could well be a warning that this amazing bull market is finally approaching its end. On the other hand, all technicians realize (or should) that all we deal with in technical analysis is probabilities.
There is, of course, the chance that the pattern this time around could be a non-event. We should point out, however, that we also have calculated some time cycles that suggest the possibility of an important market top in this time period. We are including an excerpt from our March 27, 1998 issue of Stockmarket Cycles explaining the cycle outlook for this particular cycle:
"Each decade since at least the mid-60s has had an equally spaced top of importance for that decade, and while one could argue about the relative importance of the 1973 Dow top or the 1976 Dow top, we believe we have delineated three of the most important tops of each of the past three decades and we have discovered that for all practical purposes they are equally spaced in time. In fact, they are just over a decade apart. The graph below should present the cyclic pattern to you quite clearly." Chart depicting the 10 year Cycle in DJIA.
In each of the prior three time periods, there were actually two dates that qualified as legitimate tops or double tops. We used the two tops from each time period to measure the length of the cycle in different combinations. There were some variances, of course, but the consistency between the three prior topping resolutions was notable. For example, notice that the distance between A-2 and B-1 in calendar days is 3878 and the distance between B-2 and C-1 is 3886 calendar days, a difference of only eight calendar days. Also notice the distance between A-2 and B-2 is 2721 market days and the distance between B-2 and C-2 is 2717 days, a difference of only four days between the two cycles.
The exact dates for all the points are given on the chart and the distances between points is counted in both market days and calendar days, distinguished by an "m" or a "c" at the end of each number. The third column at the bottom of the chart simply uses the averages of the two prior analogous patterns to estimate the time period of the upcoming resolution, and the dates given after the average numbers are calculations of the ideal resolution date in the current time period. For example A-1 to B-1 is 2666 market days and B-1 to C-1 is 2760 market days. The average of those two periods is 2713 market days. If we add 2713 market days to C-1, August 25, 1987, we arrive at May 19, 1998.
If we take the calculations one step further, we can use all the potential resolution dates given for D-1 and get an average date. In this case, the four possible dates for D-1 average out to May 19, 1998 with a range of April 28 to June 12. The second high, using the same technique, is due July 29, 1998 with a range from July 7 to August 22.
Finally, as a test of consistency, there is another alternative that is not explained on the page one graphic, but all the information you need is within the graphic. The technique involves measuring from A-1 and A-2 to C1 and C-2, skipping the B cycle top in the middle. After calculating those distances, you add them on to the B-1 and B-2 tops in 1977 to estimate the 1998 tops. We will not go into the details here, but we will tell you the results for D-1 average out to May 15, 1998 and the D-2 points average out to June 22, 1998.
We believe we can make an argument that, based on all the above calculations, the market is approaching an important top within the next 1-5 months with a focus on the May 19 period for point D-1 and on the July 29 period for D-2. " The Sing of the Bear" on April 6, 1998 has issued a warning. A final top could come at any time within six weeks of that signal. Seeing two back-to-back "Signs of the Bear" would be a fitting way to end the greatest bull market in our history. Be aware, however, that "The Sign of the Bear" has been given in this time period. It is a rare formation and it warns us that a signal that has preceded four of the most important tops of this century has again been seen. It does not mean the bull market must end. There have been false signals in the past. It does mean that a prudent investor should be prepared for the possibility of a significant market decline. Forewarned is forearmed.
Content copyright 2016. STOCKMARKETCYCLES.COM. All rights reserved.