Sunday, June 25, 2017

Unmasking the VooDoo - Head-and-Shoulders

Whenever I think about the famous - or infamous - head-and-shoulders pattern from the world of chart reading I always seem to go back to the late, great Mark Haines of CNBC who in the summer of 2009 asked every guest, whether they were a technical analyst or not, if they were worried about the head-and-shoulders pattern that formed in the market. I cannot remember if he was looking at the Dow or the Spoos but that is far too picky for this episode.

Anyway, what he saw was this:

It sure looked like a head-and-shoulders, which as "everyone knows" is a reversal pattern. Keep in mind the environment we were in at the time. The Internet bubble popped and the raging bear market was only three or so months in the rear-view mirror. That is, for the few who recognized that it was actually over.

Haines and countless others were still in panic mode and in extremely skeptical of any rebound. Therefore, a possible reversal pattern was looming to squash investors one more time.

I also have to give Mark credit for recognizing the pattern with it variation of a left shoulder but we'll get more into the details of the pattern a little later.

Anyway, let's see what happened after the gentlemanly freak out on television:

Oh snap! What the heck happened? The world's most recognized pattern (by non technical analysts) failed to end the rebound. And the market went up - a lot. And fast. Hokey Smokes! (Think Rocket J. Squirrel).

Let's dig into the deets, starting with what exactly a head-and-shoulders (H/S) is supposed to be.

While the pattern works in bull and bear trends, albeit it must be upside down for the latter, let's stick to everyone's favorite - the rising trend.  

As we know, as trends trend they usually exhibit a bit of ebb and flow. Advance and pullback. Three steps forward and one step back.  You get the picture. It is good when each push up makes a higher high and each fall back makes a higher low.  

The fun happens when one of those pullbacks does not make a higher low. It can make the same low as the previous low or (shudder) it can make a slightly lower low. 

Fortunately, the market or stock or commodity or bitcoin heads back up.....but cannot get back to the previous high. What we have here is a warning. And dollars to doughnuts I bet that momentum readings or volume or some other indicator makes a lower high, as well. Actually, I'll bet it made a lower high when price made a higher high, too.

The trend may be in trouble here. But is it a reversal pattern? Well, not yet.  What has to happen to complete the H/S is a lower low below the level of the previous low, or in most cases the previous two lows since they were the roughly equal. 

What may also happen at the same time is a breakdown through a significant moving average, perhaps the 50-day.  And depending on how long the trend has been in place, we may also see a breakdown below the trendline that guided the rally the whole time.

You would think that would make a compelling argument, right? And it does.

But in technical analysis everything us open for discussion. Everything. A rally can turn around without any warning at all. A stock can break out from a trading range and then break down and then break out again. We operate with the odds, not with crystal balls. Iron, maybe but never crystal.

If the odds that a stock will rally after a solid breakout from a solid pattern are 95% (I made that up) that still leaves room for failure. It is our job as chart readers to recognize failure sooner rather than later, cut losses and live to trade another day. 
Nobody gets it like they want it to be
Nobody hands you any guarantee
- Jackson Browne, Boulevard (1980)

As we can see in the second chart, the market did indeed break down from the H/S pattern. I bet the folks at CNBC were pulling spare tighty whities out of their desk drawers. 

But the darnedest thing happened. The market turned right around and started to move higher - for the next seven months with nary a hiccup. Talk about damaging your shorts!

I saw two things back then. First, basic risk control and common sense said that four trading days after the breakdown - when the market surged higher and back above the broken support line - aka the neck line - the whole reversal thing was destroyed. And if you missed that signal, certainly you abandoned your short trade two days later when the market surged again. 

If you did not see that then you really should just give Vanguard a call and let Bogle manage your stash. 

The sentiment thing was a bit trickier. People were still crazy bearish or at least crazy nervous. And the publicity given this pattern seemed rather panicky. What a lot of people don't get is that sentiment - at its extremes - tells you exactly the opposite of what is likely to happen. If everyone is bearish then there is nobody left to get bearish. There will be nobody left to sell since theoretically everyone has already sold. Supply dries up.

It's like dry kindling just waiting for a spark of demand to appear to light things up/

Sorry, got off track with a discussion of supply and demand in the markets. Anyway, the breakdown failed and everyone came rushing back in to buy.  Up, up and away!

 Let's examine the H/S pattern in 2009 a little closer: 

The pattern is supposed to be a reversal pattern and therefore it must have a trend to reverse. Typically, I like to see it last no more than one third of the time the trend lasted before the pattern started. And, I don't want it to be more than a third of the price gain of the rally.

If the H/S pattern exceeds either of these two parameters it is not a H/S reversal pattern. It may look like it but it is far too big on a relative basis so it must be something else. It may send the market lower but that's not the point. We are examining the specific head-and-shoulders pattern here. 

What we see here is that the H/S pattern was about a third of the height of the rally that preceded it. I used the Fibonacci extension ratio to get 38.2% (almost). That's close enough for me. 

However, when we get to time the patterns falls apart. The rally lasted 40 days before the pattern and you are correct, we cannot tell that until well into the pattern's development. 

The next 40-day mark is drawn in for your viewing pleasure. As we can see, the pattern lasted longer than the rally it was supposed to reverse. that means it really cannot be a reversal pattern at all. Again, it may lead to lower prices but that would be due to a simple support break. 

Well, there you have it. The H/S pattern must be commensurate with the rally it is supposed to be reversing. Sentiment was a bonus here.

And what about falling trends? Just flip the whole flipping thing upside down. Same rules apply.  We can that an inverted head-and-shoulders. Or, you can call it an upside down head-and-shoulders. Or, you can call it Ray, or you can call it Jay but you doesn't hasta call it Johnson (yeah, look up that pop culture reference). 

And now, for the pièce de résistance - the inverse or continuation head-and-shoulders. 

Nah, I lied. I'll save that for another day but here's a tip, the same rules apply in terms of price and time. 

Nothing is more bullish than a failed bearish signal.
- Mike Epstein