Last week I proposed a methodology for defining market tops in Channel Surfing and in the follow-up, Update: Channel Surfing.  I described it like this:


…characterized by a multi-month pattern within a rising market that has at least two significant touches (of the index or its Bollinger Band) of at least 1.5 standard deviations on the upper and lower extremes of a regression channel commencing after a post-correction new high.  It’s capped off by a third touch on the lower boundary and subsequent return to at least the midline before a final plunge to new lows. 

In playing around with the charts this weekend, I discovered another way of looking at these things.  First, look at the SPX over the long-term — something I don’t do often enough.

Notice how the market has climbed pretty steadily since the depression (logarithmic scale.)  It frequently departs, rising up from the trendline, but always falls back to the long-term slope.  We see five touches (1932, 1944, 1949, 1974 and 1982) each marking the end of a significant correction/crash.

Consider the first such departure: 1932 – 1942.   Follow along and we’ll play with some trendlines.

 Here, I’ve drawn three trendlines (A, B & C) coming up from the period low.  Each is drawn to a significant swing low, as is easily seen.  These are called fan lines.  You can probably see other spots that might justify one, but let’s go with these for now.  Notice how they act as support in two or more spots.  You can see that once the index finally drops through, these lines often act as resistance — they limit any further rebound.

Next, we’ll add a few fan lines dropping down from the period high (D, E, F & G.)  Notice how they act opposite to the fan lines from down below.  That is, they act as resistance until the index drops through; they then act as support.

Notice how many of the fan lines act as support to the index at one time, then turn around and act as resistance later on.  Now, let’s go crazy with them, and find all the other ones that seem “significant.”  One such line would be between D and E.  It’s an important low, although we wouldn’t know that until later.  I’ve drawn in a handful that kind of pop out.

I’ve highlighted the areas of increased consolidation.  Note that they’re all contained within one or two fan line arcs:  1 & 2, 3 & A, 4 & 5 from the low,  6 & 7, 8 & F, 10 & G from the high.  I’ve highlighted these arcs and erased the rest.  I’ve also added a few horizontal trendlines that segment and capture horizontal support and resistance.  

I see a lot of applications and implications.  But, first,  let’s focus on the topping pattern — since that’s the question of the day.  I’ve selected the 1935 – 1937 pattern.

The pattern, seen here, looks a lot like the 2000 and 20007 regression channels, right?  Enters at the midpoint, two high touches, two low touches, a final rise to the midpoint, then goodnight Irene.   But, notice that the “regression channel” lines are actually trendlines that were drawn from significant points on the previous tops.

The low line is connected to the previous high.  The high line is connected to the most significant high before that one.  The midline connects the apex of the high and low lines with the entry and exit points for the entire topping pattern.  The lower line has 5 significant touches; the top line has 3.

Hmmm, where have I seen that before?  Let’s skip ahead a few years.

The 2000 top, which I referred to as a regression channel before, could also be described as contained within 3 fan lines.  The bottom line connects the pattern’s lowest points with the previous significant high — just like in 1936.  The midline and the top line have nothing to connect to, but if I run them out to the left they intersect nicely, forming an apex with the bottom line.

The 2007 top does the same.  The bottom fan line connects the pattern’s low points with the previous significant high, just like 2000 and 1936.  The top line connects the pattern’s high points to the next most significant high.  The midline, again, forms an apex with the both of them after they connect with significant highs in 2002, and significant lows in 2001 and 1998.  Hmmm…

That brings us to 2011.  Notice the bottom fan line that connects the pattern’s low points with the previous significant high.  The top line doesn’t have any other highs to connect to, but does extend out to a few very significant lows in 2002 and 2003.  And, the midline connects nicely to the apex.

The astute reader (with way too much time on their hands) might recall I pondered whether the current top was fully formed, or might expand as happened in 1999.  The first 3-4 months of that top looked very similar to the past several months of this pattern, then widened to a pattern that lasted another two years (see the 4th and 5th charts in Update: Channel Surfing.)

I worried that the current top wasn’t large enough in size or time compared with the previous tops or with the previous summer 2010 correction.   With this new understanding, it’s easy to see how and why the top patterns expand.  The 100 point range of the past month would grow to 125 points by the middle of August [Sure it Works in Practice] and 165 points by the end of the year.

So, is that what’s happening?   Trust me, I’m working on it.

In the meantime, look again at the 2000 and 2007 tops.  Thanks to my fledgling understanding of fan lines,  it’s pretty clear to me that many of the patterns in the current topping pattern have been driven by fan lines from the 2007 top and the 2009 bottom.

I drew the fan line from the 2007 top [The Trendline That Just Won’t Quit],  the trendline from the Mar ’09 bottom, the midline of this topping pattern, a trendline off the Nov 30 and Mar 16 lows and a trendline from the May 2 and Jun 1 highs.

Interestingly, they all intersect around June 23 at about 1320.

Now, I don’t know whether we’re in P[3] or Minor 4. I don’t know whether to be more worried about the Puetz window or Three Peaks and a Domed House or the Egg of Doom.  Those are issues for another post.  But, I have a pretty strong hunch that we’ll bump back up from these levels before anything else happens.

Stay tuned.

UPDATE:  June 17, 2011

The pattern indicated an upside to 97.42, which would be a possible turning point.  TLT closed yesterday at 97.38 and was as low as 96.70 this morning.  Let’s keep an eye on this one, if for no other reason than as an indicator.

ORIGINAL POST:  June 15, 2011

Posted yesterday midday, TLT was around 95.4 at the time.

“For anyone trading TLT, it completed a pretty nice bullish butterfly at 95.5, indicates an upside of 96.67 at .618, 97.90 at 1.272 and 98.55 at 1.618 extensions. “

Today, TLT is pushing 97, although at 97.42, it could start looking like a candidate to go the other way in a bearish pattern.

But, as so often happens, the completion of a bearish pattern is the start of a bullish pattern, and vice versa.  If it reverses here or at 97.40, watch it as it approaches 94.83.  A stall there would be a potential setup for a bullish butterfly.

From an equity standpoint, TLT is moving opposite SPX these days.  My bias is that we’re due for a rebound of more than 60 points in SPX; it would most likely arrive in an A-B-C pattern.  Hence, the above scenario in TLT could make a lot of sense.

I know… another round of horrid economic news, escalating unrest in Greece, etc., etc…

Yes, the economy is still melting down.  No, there’s nothing we can do about it.   There’s still no miracle cure, no silver bullet and no pain-free long-term solution.  But, that doesn’t mean they won’t try.

The two leading options right now are that we’re in a triangle ( expanding or running flat), trying to finish out minor 4 and launch 5 of C of P[2], or we’re in [iii] of minor 1 of P[3].  Yesterday, the triangle felt right.  Today, P[3].  I’ll talk later about which I’m expecting and why.

But, either way, we should see more of a bounce than we got yesterday.  I’m calling this (1261.90) the bottom for now, looking for a rebound starting tomorrow.  My initial target is 1328 on June 21-23.

If the SPX slide stops around here (ideally 1263), we would have a butterfly pattern (since the open, visible on the 5 minute) nestled inside another butterfly (also on 5 minute, since Friday’s low) nestled inside a crab (since 4/18 on the daily), nestled inside yet another crab (since 3/16 on daily.)  Think of them as four very bullish Russian nesting dolls, collectively indicating an upside of 1329 or better.

Incidentally, 1329 is the midline of our 2-std dev regression channel we’ve been in for the past 4 months.  If this is the top, we always retrace to the midline after the 2nd touch of the negative 2 std dev line one last time before the plunge.  http://pebblewriter.blogspot.c…

Not saying it has to happen… but VIX seems to be contained within the same channel it’s been in since April, and whatever the wave count, this isn’t feeling like a meltdown to me — just the rebound camp getting a little ahead of itself this morning.

Here are the patterns.  The butterfly patterns are nestled at the end of the CD leg in the lower right corner.

Two Crab Patterns

Two Butterfly Patterns

Early results from reviewing a lot of market tops and would-be market tops…  I have yet to find an example of the model I suggested not working.  And, folks, this is kinda exciting stuff. 

To recap, about a week ago I noticed [Watch for the Rebound] that the drop we were experiencing might set up an upward-trending channel that would embolden the bulls.  The channel could be seen by drawing a trendline off the 1344 and 1370 highs, and a parallel bottom between the 1249 low and the to-be-established low. 

By this past Friday, it dawned on me [Channel Surfing] that what I had seen was actually a two-standard deviation regression channel.  The tops in 2000 and 2007 were very similar in the way that they entered into and behaved while within such a channel.  More importantly, the current market has behaved very similarly to those tops.

I spent a few hours today dragging two-standard deviation regression channels around to various market tops and what looked like could be market tops.  Bottom line,  I’m fairly well convinced that not only have market tops all behaved similarly, but that the pattern observed is predictive of a market top.  In fact, I’m leaning towards calling this pattern a requirement of significant market tops.

It’s characterized by a multi-month pattern within a rising market that has at least two significant touches (of the index or its Bollinger Band) of at least 1.5 standard deviations on the upper and lower extremes of a regression channel commencing after a post-correction new high.  It’s capped off by a third touch on the lower boundary and subsequent return to at least the midline before a final plunge to new lows.  Here’s the view of the 2000 and 2007 tops I posted last week.

2000 TOP

2007 TOP

And, here’s where the market is now:

2011 TOP?

 

So, while I’m pretty confident about the next move (up, in the short term), is this THE top?

Even though it’s predictive of a top, this pattern doesn’t in and of itself mean a major top is imminent.  Consider early 1999.  From March through May, the pattern displayed perfectly.  It featured a 100 point drop from May 13 through May 27, including 3 up and 3 down days of 20+ points off a 1375 high (hmmm… sounds familiar.)

WHAT MIGHT HAVE BEEN

Sure enough, after the 3rd touch of the bottom, the index rebounded to the midline at 1336.  As might be expected, it fell back — but only to 1287, from where it traced out five waves up over the next month to 1420!

WHAT WAS

In a turn of events that only a fractologist could love, it retraced the entire pattern on a much larger scale over the next 1 1/2 years before finally resulting in the crash we all knew and loved.  The initial, smaller scale channel can be seen in the far left of the much larger/longer pattern in the graph above.

What’s an investor to do?  Look at the market’s behavior leading into the supposed top.  In 1999, the market had just experienced a 12-day, 180 point swoon.  So, a topping pattern that spanned only 100 points or so wasn’t at the same scale.

In 2007, the last major correction was just over 100 points (May – August ’06.)  So, a topping pattern that spanned 200 or so points was completely in scale with the preceding action.

What about today’s market?   So far, this channel is about 100 points.  We had a 50-point, 1-month correction last November, so that seems reasonable.  But, look back at the summer of 2010.  We had a 200-point correction that lasted 7 months.  The current topping pattern seems a little on the small side by comparison.

And, what’s up with the November pattern looking suspiciously like the summer pattern?  Same thing happened with that fakeout in 1999 — a 65-point droop (also visible above) two months before the fun started that’s the spitting image of the 180-point one a few months before that.

Besides, look at the time involved in the downturns preceding the pattern.  In each case, the larger correction was less than half the time span of the ultimate topping pattern.  Last summer’s correction lasted 7 months, but this topping pattern is only 4 months old.

Could this be a head-fake, too?  Short answer: yes.  I have little doubt that we’re eventually going down after this week’s bounce back to the channel midline.  My inclination is that it’s sooner than later.  After all, we did lose the trendline of support that’s kept us going since Mar ’09 and the rising wedge since Oct ’10.  The economy is on life support.  And, ending QE2 is going to feel like cold turkey to this addicted market.

But, just to be safe, I’ll wait for a breakout in one direction or the other to tell me whether or not a larger pattern is developing.  While I think it’s 50-50 at best, the possibility remains that this is a Minor 4 triangle, preparing the way for a Minor 5 push to 1370 or more.

Upon nearing the midline, I’ll place stops to protect the downside, and not go short until/unless we break below the channel.  If the past is any guide (and isn’t it always?) the market would come back and backtest the channel before heading further south.  So, at most, it’s an opportunity loss of 50 points or so.  If this really is P[3], I’ll never miss those 50 points.

I know what you’re thinking:  this is diametrically opposed to last week’s forecast (which seemed so great at the time!)  But, given that the original 87-day cycle date isn’t until August 11th, it would make sense if the plonger énorme were delayed a bit.  And, the leading candidate for the return to the midline isn’t until 6/29 or so.  We could bounce around a bit, I suppose.  But, how would we kill another six weeks if it isn’t by tracing out a (no doubt truncated) Minor 5?


Just for the record, I think this market stinks.  I’m bearish.  Period.  There’s nothing happening in the world that gives me even the slightest confidence that we’ll come out of this looking good (okay, maybe the implosion of Newt’s campaign.)  Any move up, if it happens, would be the direct result of market manipulation on the grandest scale by the Fed — which, of course, is desperate to prevent P[3].  Would they do it?  In a heart beat.  Could they do it?  Maybe — at least temporarily.  Lots of economic reports coming out next week, most of which could be pushed one way or the other.  

Or, given that the war is essentially lost anyway, why not blow the remaining ammunition on a last gasp effort to turn things around and announce QE3?  It would ultimately do more harm than good,  but it might buy some time.  Maybe call it something else, so people don’t catch on right away.  Qualitative Easing, anyone?

Stay tuned.

Just a quick update on yesterday’s post [Deja Vu].  Today was b’tugly from the get go, with the SPX down below 1270 at its worst.  Despite all that, the channel remains intact.

Not content to see that something works without understanding how, I went back to the 2000 and 2007 tops.  Made a fascinating discovery.  The channels I’d drawn, based on what the charts showed me, actually work out to be the trendlines drawn out 2 standard deviations in a regression channel.  Check it out:

2000 TOP

TOS draws these things for you, so it’s an easy matter to set the lines at -2, -1, 0, 1 and 2 std dev’s and let it show you the way.  As would be expected, the +/- 2 std dev trendlines (outside, in red) line up with the bollinger bands.  The entry point for the channel is when the index hits the mid-line.  Its first retracement marks the lower end of the channel, and its subsequent high(s) marks the high end.

Throughout the pattern, the mid-line acts as a pretty good support/resistance line on an intermediate basis.  And, of course, the +/- 1 std dev lines capture the bulk of the advances and declines (by definition.)

What’s really interesting to me right now, though, is what happens at the end of the pattern.  When the index loses its long-term trendline (e.g. falls out of the rising wedge) and all seems lost [watched CNBC today?], the index bounces off the lower end of the channel one last time and makes a return trip to the midpoint of the channel.  Happened in both 2000 and 2007.

2007 TOP

Also note that the index, when it seems done with the channel bottom, still comes back for one last touch before heading south for the winter.  So, if you’re sitting around, wondering where the panic is if this is THE END, now you know why.  As b’tugly as things look, this ol’ gal has one last dance left in her.

Here’s where we are as of late Friday afternoon.

2011 TOP

If the pattern repeats, it gives us a little more precise idea of there the market should turn.  It appears to be at 1328-1330, where the central line of the channel crosses the bottom of the rising wedge.  I’ll spend the weekend looking at other tops just to double check.  So far, more than a few pass the eyeball test, but I’d like to make sure. 

Have a great weekend, everyone!

I’ve been looking at the tops in 2000 and 2007, trying to draw comparisons.  There are similarities in the way the market, once it drops below its long-term support (bottom of a rising wedge), looks like it’s in for a free fall.

The 2000, 2007 and 2011 Tops Weekly

But, it eventually finds new support in a parallel channel, as I theorized a few days ago [Watch for the Rebound].  The bottom of the channel is drawn off a recent major high and the top is drawn off the two most recent peaks.  It pencils in nicely for 2000 and 2007, although one could argue for a different placement, depending on the “degree” being examined.

In any case, if SPX bounces as I expect here at 1280, it will have set up a similar channel.  The implication is that we will, indeed, start what looks like a new bullish move to the upside.

It’s not.  Judging from past results, it’s a major headfake that will retest the former support line (bottom of the rising wedge) before resuming its decline. Where?

Forecast alert!

I expect the rebound to stop somewhere short of 1320 before the end of June — ideally Wednesday the 29th.  It’s the intersection of the rising wedge, a trendline drawn off the May 2 top and the fan line from the Oct ’07 high [the Trendline That Just Won’t Quit.]  And, here’s the obligatory mention of the end of QE2.  Once we start down, we’ll complete the H&S; that’s been forming for six months (the channel bottom is the neckline).

Our first significant target is the Nov ’10 highs around 1225.  We should get a bounce there that takes us back up to test the bottom of the channel somewhere between the middle of July and the middle of August [Sure It Works in Practice] before resuming the downturn.

Bottom line:  we have about two weeks left to plan those masterful bearish trades, but there are still a couple of pennies left in the path of this steamroller.

2000 Weekly

2007 Weekly

2011 Weekly

2000 Daily

2007 Daily

2011 Daily and Forecast

I’ve been testing various methods of assessing market behavior, trying to make some sense of where we’ve been and where we’re going.

I’m increasingly partial to the Fan Principle, discussed by John Murphy in Technical Analysis of the Financial Markets (p. 74).  The basic premise is that support lines act as such until they’re broken.  They then become resistance lines that contain the ensuing rally (backtest) until the next lower level support line is established.  Once broken, that new support line becomes resistance, etc. etc. 

Most practitioners believe that after breaking the 3rd fan line, the market is heading lower (than the original pivot point, at least) — kind of a three strikes and you’re out approach. 

It’s pretty easy to see over the past 25-30 years.  The S&P; 500 is cruising along until 1994, when it suddenly develops an extremely bullish hair.  It triples in 5 years, soaring to over 1500 in 2000 before falling back to 945 by Sep ’01.  I draw trendline A to this point.

It reverses here, gaining over 200 points before running out of steam and falling back to TL-A.  This time, however, it plunges through, shedding another 250 points on its way to 769 in Oct ’02.   I draw another line here (TL-B). The market then backtests TL-A, now resistance, eventually giving up in Oct ’07.

It plunges through TL-B, and goes in search of TL-C, finding it in Mar ’09 at 666.  Since then, we’ve been in backtest mode, trying to claw our way back to TL-B.  Pretty simple, right?

There are plenty of details to consider.  For one, which peaks and troughs are major enough to be pivot points for one of the trendlines?

First, the Trees…

Here’s the updated table to reflect yesterday’s action.  Please note the “Days 1-3” data reflects the high price during the first three days after each decline, regardless of which day it falls on.

I also included a Days 1-10 column to reflect the high price during the first ten days; it also shows the number of trading days (post decline) required to produce that high.

A couple of interesting things popped out of the additional study.   First, five or more declining days in a row is even more rare, occurring only 5 times since the Oct 07 highs (now 6).   Previous 5th day declines were 2, 5, 8, 15 and 61 points, so yesterday’s 1.37 was the lowest yet.

Now, of those five previous 5-day declines, only three went on to decline a 6th day.  They’re seen in the table as the periods ending 10/10/08, 2/23/09 and 7/2/10.  The first two occurred in the midst of the biggest bear market in years.  The next few days in the 10/10 period were -4, -78, -3, +90 and up an intraday 44 on the last day before closing down again.  The 2/23 period resulted in a 40 point drop on the 6th day, followed by a 28 point gain on the 7th.  Both these periods hit an intraday high on the 2nd post-decline day before resuming their bearish trend.

The 6th day in the 7/2/10 period started off down 4, raced back up 14 points, then fell back to close down 3 cents.  But, the next five days saw a rebound that retraced the entire decline and then some (140%).

There were two other declines that ended after 5 days.  The 1/22/08 period retraced 29%, 58% and 85% in the 1-day, 3-day and 10-day periods respectively.  The peak was on the 8th day.  The 7/2/10 period retraced 23%, 23% and 111% and peaked on the 10th day.

In general, 2nd and 8th days were important in terms of retracement highs.   They accounted for over half of all the periods.  An intraday high frequently occurred on the 2nd day post-decline when the market was tanking.  Retracements following declines in upward-trending markets usually benefited from the extra days of market advances.

Now, if you haven’t dozed off yet, the really cool stuff.

Finally, the Forest…

The @ symbol marks each of the 4+ consecutive day declines since the 2007 highs.  See anything interesting?

If the reported settlement of $20 billion is all it takes to get the banks out of their foreclosure fraud liability, we should see a pop in the financials.

I’ve been watching a bullish Bat pattern evolve in XLF.  While I’m very bearish on the financials (and the market in general), a settlement could provide a short-term boost that leads the XLF and the overall market higher over the next few days.

The pattern targets an upside of 16.55, but a return to the upper end of the channel is likely all we’ll see.

One note of caution for traders: this pattern’s CD leg is currently approaching a 2.00 extension of the BC (1.94 at today’s low.)  Bats can also extend to 2.24 or 2.618, which would result in continued downside to 14.6 or 14.24 respectively.

That would drop XLF out of the channel, which is entirely possible.  Experienced harmonics traders look for a confirmed move in the anticipated direction before placing trades, and place stops to limit the fallout from the 30% failure rate.