Deadly Divergences (Part 1)
10/09/2008 1:18 pm EST
Though it is sometimes hard to admit, I have been writing about divergence analysis for over 20 years. I have noticed in the past few years a common theme in articles about divergence analysis that I feel does not allow it to be judged correctly as a technical tool.
In many articles, whether they are trying to prove that divergence analysis works or does not work, they only look at one time frame. I have always looked at divergence analysis using two different time frames. From my early years, this has meant that I first look at the weekly data and if the indicator is confirming the price action by making higher highs or lower lows, then I gauge the intermediate trend to be intact. In these examples, I am using the 14-period RSI with a 21-period weighted moving average (WMA). It should also be mentioned that in very strong up or downtrends you may see the formation of four to eight week divergences that occur at neutral levels, not overbought or oversold territory. These often warn of corrections within the intermediate-term trend. Typically, at major turning points, you will be able to observe the formation of one or more divergences at overbought or oversold levels that form over a six to 20 week time period. These divergences then must be confirmed by a break below significant support or above resistance. When daily negative divergences are identified but no divergences exist on the weekly data these daily divergences just signal corrections within the intermediate-term trend. Some analysts who are just looking at divergences in one time frame conclude that divergence analysis does not work. I would disagree with this, but it has always been my view that before you use a method of analysis, you must first check it out yourself; otherwise, you will not have the confidence to use it when it counts.
Let's look at the weekly analysis first. I have chosen Apple (AAPL) one of the most popular and widely followed stocks of the past few years. This chart covers the period from March of 2007 through early September of 2008. In April of 2007 the RSI moved above its 21 WMA (green line) as AAPL started its rally from the $90 area. Both the price and RSI continued to move higher until July when the iPhone was released. The stock peaked just below $150 with the RSI at 83. After a three-week correction that took AAPL down 25% from its highs, the stock once again turned higher. On this correction, the RSI dropped back to the 55 area, which I have observed is quite normal. By the middle of September, both the RSI and prices were moving higher with AAPL exceeding the $190 level in early November. The RSI failed to confirm these highs, line A, as it peaked just above 81. After a two-week correction AAPL again moved higher and by late December, surpassed the $200 level. At these highs, a secondary negative divergence was evident in the RSI, line B, as it peaked in the 72 area. This suggested that an important top might be in place. Two weeks after the highs, line 1, the RSI dropped below key support at line C. Over the next six weeks, AAPL dropped almost 30% before stabilizing in the $115 area.
By the end of March with the close at $143, the RSI moved back above its WMA, and AAPL rallied slightly above $190 over the ensuing seven weeks. The weekly RSI rallied up to the 65-70 area, and by the latter part of June, had dropped back below its WMA. After a six-week decline the RSI again bounced and just made it slightly above its WMA. The violation of the seven-month RSI support, line D, was a strong signal that the downtrend had resumed. Now let's examine the daily chart.
The daily chart above begins in March of 2007 and goes through April of 2008. On April 28, (blue box) the weekly RSI moved above its WMA, and the weekly triangle formation was completed with the close above $99. The daily RSI was already above its WMA, therefore, the weekly and daily trend were both positive. The daily RSI peaked a month later and then as AAPL continued to move higher, the RSI formed a series of three lower highs. The lower highs occurred very close together which made them less significant and with no divergences in the weekly analysis these divergences were consistent with either a correction or a period of consolidation. After peaking on June 7 just above $125, AAPL formed a triangle over the next four weeks. The daily RSI confirmed the completion of the continuation pattern as the RSI again moved above its WMA. The rally continued into July, but the daily RSI formed a longer-term divergence at these highs as indicated by line a. The weekly RSI had not yet formed any bearish divergences, as highlighted by the green box, so once again the daily divergence were indicating a correction not an intermediate-term top. However, the length of the daily divergence suggested this correction might be more significant than the prior correction. The decline retraced just over 50% of the rally from the December 2006 lows at $76.77. It should also be pointed out that decline took place as the overall market was dropping sharply. During this decline, the weekly RSI dropped back below its WMA which increased the odds of a weekly divergence forming at the next new highs.
Just a few days after the lows the daily RSI was back above its WMA, but it took until the end of September for the weekly RSI to move above its WMA. Still, the next rally was dramatic as AAPL made a high of $192.68 on November 7 (line 1) but then closed lower for the day and four days latter made a low of $150.73. The November highs were not confirmed by the weekly RSI (Figure 1), and on the daily chart this period is highlighted by the red box. The daily RSI also did not move above the May highs. As AAPL began to rally once more and the daily RSI moved back above its WMA but was much weaker adding to the evidence that an intermediate-term top was forming. The price action was impressive as AAPL hit a high of $202.96 on December 27 but the weekly and daily RSI were both lagging. The weekly RSI formed a second bearish divergence (Figure 1) and just barely made it above its WMA, and the bounce in the daily RSI was also anemic. Therefore, the break of weekly RSI support in early 2008 (line 2) turned both the weekly and daily trends negative setting the stage for the 30% decline. The break of the uptrend in the daily RSI, line c, was also negative. AAPL did make a nice secondary peak just above $192 in May of 2008 but as noted early it was technically weak.
The chart above is on Transocean Inc. (RIG) with the weekly chart covering November 2005 through September 2008 while the daily chart spans December 2007 through September 2008. RIG is a good example of how the weekly can give you a much clearer picture of the market action than the daily analysis. In early April of 2007, RIG completed a weekly triangle formation (see arrow) that was confirmed by a move in the RSI above the year long resistance, line a. The several month rally took RIG from $87.40 up to the $117 level in late July 2007. The RSI peaked above 81, point 1. After a quick three-week correction, RIG resumed its uptrend, reaching the $150 level in late December 2007, point 2. The weekly RSI did not confirm these highs, point 2. Two weeks later, the uptrend in the RSI, line b, was broken. The RSI then dropped back to the 50 level and then again tried to turn higher. Even though the RIG made significant new highs above $160 in May 2008, the RSI formed another bearish divergence at point 3. These divergences were then confirmed by the break of the 18-month RSI support, line c, the week of July 18, 2008 (line 4). The weekly RSI has continued to drop sharply along with prices and has stayed well below its declining WMA. The signals from the daily studies were consistent but not nearly as clear. The daily RSI formed higher highs in February and April but then violated its uptrend, line d, on April 29. This weakness was consistent with the failure of the weekly RSI to confirm the December 2007 highs. As RIG made another new high in May 2008, the daily RSI was lower, line e. The daily RSI then formed a series of lower highs and lower lows, line f, and the weekly RSI support was broken in July, line 4. The daily RSI has subsequently moved above its declining WMA several times as some of the rebounds have been quite sharp. The lack of any positive weekly signals, however, should have kept you short or on the sidelines.
These two examples should illustrate the danger of just looking at daily RSI values as by relying only on them, you will, in most cases, get a misleading picture. Here are some guidelines that might help you to use the RSI in your analysis. Look for weekly negative divergences that form over a 6-12 month period, or longer, to identify significant market tops. These need to be confirmed by a break of support that goes back at least a year. Short-term negative divergences, especially those that are not formed at overbought levels, just signal corrections. Some of the time, the daily RSI can help you fine tune entry points when the weekly analysis is not confirming. In summary, be wary of daily positive divergences that develop in a downtrend or negative divergences in an uptrend if they are contrary to the weekly analysis. In the next part, I will apply this same type of analysis to other technical studies.