Head And Shoulders Probabilities

Posted At 9:00 AM EST

Good morning...We begin the day with some overhang of selling pressure after the heavy selling from late last week.  Composite volume was surprisingly heavy for a session immediately before a holiday break.

Surely the hottest topic among most traders, and certainly those with a technical bent, is the current Head & Shoulders pattern in the S&P 500.

If by chance you've been a lazy bum over the past week or so (I'm sheepishly raising my hand...), then perhaps you haven't heard.  Very quickly, a Head & Shoulders pattern looks like, well, a head and two shoulders.  It is commonly believed to be a topping pattern with a very clear trigger (a violation of the trendline between the two lows) and target (the distance between the violated trendline and the top of the formation, projected down from the trendline).

There are two problems with this:

1.  Just because the pattern has formed doesn't mean that it's going to trigger, and

2.  Even if it triggers, we're not guaranteed a trip down to the target area.

I'm not going to discuss the theory behind the pattern - there's nothing new I can add there.  But given the intense interest, I thought it would be interesting to look at whether these patterns usually do trigger, and when they do, if the target area is usually reached.

It's very difficult to test a pattern like this because of the multiple variables, but I did find some code from an old Technical Analysis of Stocks & Commodities magazine (April 2002).  After some tweaks, it did a decent job at highlighting these patterns.  Some were kind of goofy, but overall it was able to pick out what most would recognize as  Head & Shoulders formations.

What it looks for are times when a stock forms three peaks, with no more than two months between any of them.  There has to be at least a 4% retracement between the peaks to qualify, the left and right shoulders must peak within 1% of each other (so there's some symmetry to the pattern), and the head must be at least 1% higher than the highest shoulder.

I could only find 7 previous occurrences in the S&P 500.  Of those 7, all but one followed through by violating the neckline.  And of those 6 that triggered, 4 of them went on to meet its downside target.  During the first three days after the break of the trendline, the S&P continued to sell off 4 of the 6 times, averaging -3.6%.  The two times it rebounded in the short-term, it did so strongly, and continued to rally for months afterward, basically negating the H&S pattern.

That's not a very big sample size, so with some great help from my youngest son (I promised I'd mention him), we went through every stock in the S&P 500 over the past 15 years looking at the formations.

Here's the final tally:

• Number of H&S patterns that set up:  1,322

• Number of them that actually triggered:  870 (65% of the total)

• Number of targets reached by those that triggered:  757 (87% of the total)

The pattern had a pretty high follow-through rate, with nearly two-thirds of all setups actually triggering the pattern by the stock violating the neckline.  And of those that triggered, there was a remarkably high probability of actually hitting the projected target, with 87% of them doing so, usually within one month of the trigger.

Again, some of the patterns that the code identified were loose, and I didn't necessarily agree with how the trendlines were drawn, which threw off the projected target levels.  But overall it was in line with common technical guidelines, and does suggest that should the pattern trigger in the S&P, the 850ish area on that index will become the next target level.  About the only real exception to that would be if the index turns right around and rallies after violating the trendline.

Bottom line - Intermediate-term Outlook: Neutral (since April 9, SPX 843)

Beginning in early March, we discussed a large number of reasons to expect an imminent rally of one to three months' duration.  Some of those studies were even more positive, and suggested not just a rally, but possibly a new bull market.

During mid-April, the market held up extremely well in spite of some overbought types of indications.  This is very rare during an ongoing bear market, and is important to keep in mind especially given many of the "this time is different" kinds of studies we reiterated in early May.

While there were - and continue to be - many reasons to consider this rally something different than we'd seen previously in the bear market, I was looking for the S&P to run into trouble if it traded into 940-950, which happened early in June.  I wasn't expecting any kind of waterfall decline to new lows, just more of a pullback than we'd seen.

With the most recent surge in the spread between the Dumb Money and Smart Money Confidence, and the tendency for initial breakouts from volatility coils to be "false", I was looking for the first breakout above 950 to be beaten back.  Now that that's happened and we're nearing the opposite end of the May - June range, we need to see how the market responds to short-term oversold conditions, especially now that we've seen a "failed" rally above the 200-day average.

The recent 90% down volume readings when coming off of an intermediate-term high aren't necessarily a sign that the trend is changing, but if we can't get meaningful bounces from short-term oversold conditions early this week, and especially if we lose the 880ish area on the S&P which would trigger the Head & Shoulders pattern we discussed above, then the 850ish area will be the next focus.

Bottom line - Short-term Outlook:  Neutral (since June 30, SPX 919)

The selling pressure on Thursday finally followed through on some of the short-term negatives we'd discussed heading into last week.  It took a bit longer than expected, but the pressure was enough to cycle some of our shortest-term guides back towards or into oversold, such as the STEM.MR Model.

Since the March low, the market has done quite well at rebounding over the next 1-3 sessions when that model has reached oversold territory and began to curl higher (it hasn't started to curl yet, so it could become even more extreme).  As always, if the market cannot bounce from short-term oversold conditions, it is a sign of inherent weakness and is often a heads-up that the larger uptrend is changing.

That is especially the case when the oversold conditions occur near support, which we're on the cusp of now with that 880ish area on the S&P 500 that everyone is watching.

The last time the S&P futures lost 2% or more before a weekend, then gapped down at least -0.5% on Monday morning was on March 2nd, so obviously it's been awhile.  There have been 20 times it has happened since 1982, and usually we've seen a rebound into mid-week but not necessarily right from Monday's open.

Buying Monday's open and selling at Monday's close resulted in only 7 winning trades, for a 35% win rate.  The average return was -1.2%, so obviously there was most often some additional follow-through selling pressure into Monday's close.  But buying Monday's close and holding through Wednesday's resulted in 80% winning trades, an average return of +2.7% and an average maximum reward (+4.0%) that greatly outpaced the average maximum risk (-1.7%).

I'm leery of a bounce right off this morning's gap down, but if we happen to see some additional selling into the close, that takes us near that 880ish area on the S&P 500, I'll likely be looking for a bounce into mid-week.

All the best,