Deadly Divergences (Part Two)
12/04/2008 12:01 am EST
In the first part of this article, I explained my belief that divergence analysis must be done 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. 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. Only daily signals that are in agreement with the weekly trend should be followed. 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. In part one, I used the RSI and would like to concentrate on the OBV in part two.
First of all, let’s review how I use the OBV for those who have not read my previous articles on the subject. The weekly data is the most important to look at, and this period in the history of Hewlett Packard (HPQ) is a textbook example. Below the bar chart is the OBV with a 21-week WMA in green. The downtrend in the OBV (see line a) was broken in July 2005, and by the end of the month, was above its WMA, which had turned up. HPQ then moved from $24.50 to a high of $34.50 in February 2006, but the OBV started to diverge in early 2006 as it moved sideways (see line b). Prices corrected for the next six months with HPQ dropping back below the $30 level. During the week of August 16th, the OBV moved back above its WMA, and by the end of September, the OBV overcame resistance and was soon making new highs once again. The OBV peaked in May of 2007 (see line 1) with HPQ trading at $45.80. HPQ managed to continue higher, peaking in early November at $53.48, but the OBV failed to confirm these highs (see line 2). The divergence can be observed by comparing the slope of prices (see line d) with that of the OBV (see line e). Once a significant weekly divergence is formed, we look for a break of key support or a long-term uptrend to confirm the divergence. The two-year weekly uptrend in the OBV (see line c), was broken in early January of 2008. Since this support was broken, the OBV has been making lower highs and lower lows, staying below its declining WMA.
Of course, once the weekly analysis has been done, I normally look at the daily analysis to see if they are in agreement or not. I have found that, when in doubt, I am better off relying on the weekly, rather than the daily data. The daily chart of HPQ just covers the latter part of 2007 and early 2008. The daily OBV peaked in early October, and when HPQ was making new highs on November 6, 2007, the OBV was lower (see line b). Once again, the slopes of the price and OBV (see lines a and b, respectively) illustrate this negative divergence. By itself, this relatively short-term daily divergence would not have been significant, but with the longer-term divergence in the weekly OBV, the daily divergence was significant. HPQ reversed to the downside the next day (see arrow) and dropped almost 10% in just a few trading days. HPQ then bottomed in the $47-$48 area and turned higher in late November. On December 5th, the resistance in the OBV (see line b) was overcome. When the weekly analysis is positive, a move through bearish divergence resistance on the daily chart signals a resumption of the uptrend. In this case, the weekly OBV had already formed a significant negative divergence, which was a good reason not to buy. In fact, the daily OBV became even more misleading by the end of December as it made new all-time highs (see line 1). However, the weekly OBV was well below its previous peak. Over the next eight days, HPQ dropped sharply, losing $10 per share.
In the next example, we will look at the SPY on both a weekly and daily basis, concentrating on the late-2007 top and the decline through November of 2008. The week of October 13, 2007, the SPY closed at new weekly high of $156.30. This was not confirmed by the weekly OBV, which formed slightly lower highs (see line a). The daily OBV also did not confirm the highs (see line 3), and with the lower high in November, a downtrend in the daily OBV (see line d) was established. The weekly OBV held support (see line b) at both the January and March 2008 lows. The bear market rally from the March lows to the May highs just tested the downtrend in the daily OBV (see line d), while the weekly OBV moved slightly above its flat WMA. The violation of the weekly OBV support in June (see line 1), and the declining weekly WMA generated much stronger sell signals. On the rebound from the July lows, the weekly OBV stayed below its declining WMA, while the daily OBV broke out to the upside and spiked sharply higher at the end of September. By itself, this could have been interpreted as a very bullish signal, but the fact that the weekly OBV was just breaking support (see line 2) confirmed the intermediate-term downtrend. The weekly OBV continued to make new lows as the attempt to stabilize in October failed. The daily OBV had also confirmed the new lows, and typically we would see some positive divergences in the daily studies before a decent rally against the major downtrend.
The emerging markets have been under extreme selling pressure for quite a while, so for the next example, we will look at the MSCI Emerging Markets Index ETF (EEM). From September 2006 through July 2007 (point 1), the weekly OBV stayed above its rising WMA and confirmed the price action. On the sharp drop in the summer of 2007, the OBV dropped back below its WMA. The rally from the August lows was very powerful, as EEM rallied from a low of $37 to a high of $55 (point 2). Though the weekly OBV did move back above its WMA, it failed to confirm the new highs as it was significantly lower (point 2) The weekly OBV managed to hold support (see line a) until March 2008 when it was broken. Then, a sharp oversold rally from the March 2008 lows peaked at $52.48 the week of May 24th, but EEM reversed to close the week lower. On this rally, the weekly OBV just moved back to its declining WMA, which was negative. The subsequent break below support (see line b) reaffirmed the intermediate-term downtrend. The daily analysis gave a much different picture, as the OBV did confirm the late October 2007 highs and actually peaked in March of 2008 (see line c). Unlike the weekly OBV, the daily OBV almost made new highs in May, but soon turned lower and broke support (see line d), which was negative. By using both the weekly and daily analysis, the weekly sell signal in March should have had traders waiting on the sidelines for new sell signals from the daily studies before establishing a new position. This would have meant missing the 20% rally from the March 2008 lows to the May highs, but with the strength of the recent decline, it probably was worth it. Both the weekly and daily continue to make new lows with no signs yet of a bottom.
Though the OBV is one of my favorite indicators, it does not work all the time, and I thought it was important to also provide an example of how the OBV failed to identify a significant top. RIMM has been a widely watched stock for the past few years, and here, the weekly OBV broke through its downtrend (see line a) in September 2006, with RIMM closing at $29. After reaching the $47.50 area, RIMM consolidated for several months before starting a new leg to the upside, with the OBV moving to new highs (see line b). The OBV stayed well above its rising WMA until the $130 level was reached. Another three-month correction followed before RIMM made a push to its all-time highs at $148. The OBV continued to confirm the new highs and held up well on the decline into the July 2008 lows. The weekly chart support was broken two weeks before the OBV support (see line d), and even though RIMM had dropped almost 70% from its highs, the long-term uptrend in the OBV (see line c) was not broken until October 2008.Of course, all traders should be aware that any strategy needs a disciplined money management component to be successful, and this should have gotten you out of RIMM before the plunge got really nasty. I hope this two-part article has convinced you that using divergence analysis on only one time frame is not a good idea, despite what you may read in some popular publications. Therefore, if you see a good setup on the daily charts, be sure to check the weekly charts before taking the trade, as this may keep you out of some losing trades.