Sunday, January 23, 2011

Stepping Back for the Macro View

.
I always consider myself fortunate when I have the opportunity to stand in front of folks more experienced than I at this thing we call trading/investing.  Even though I was on the podium, I always feel that I learn more from the membership than I impart, and I am thankful for the time and attention paid by my audience.

Thank you for the wonderful support shown yesterday by the local AAII Computerized Investing group here in NoVA -- and specifically to Bob Bigrigg for the invite.

==============

I always have to step back and review the bigger picture when I'm preparing for a presentation, and in doing so, I'm always, without exception, playing catchup to the fundamentals of my process.  My fundamentals are quite simple:

  • Big Picture Analysis
  • Opportunity Analysis
The idea of standing in front of people, most of whom are far more skilled than I, puts a fear of the unknown in me and forces me to be at the top of my game.  Whether I actually *was* at the top of my game is a subjective answer that eludes positive identification, but nevertheless, the preparation always makes me more confident in my convictions and is the equivalent of "sharpening the saw".

Let's start with the Big Picture.

=============

The Big Picture

I rely almost exclusively on my GGT system for information.  Recall that the stock database is comprised of what I think is a baseline of quality stocks:
  • The stocks must have a share price of greater than $1, determined weekly at the close of Friday's market;
  • The stocks must have a volume of greater than 100,000 shares, again determined weekly at the close of Friday's market;
  • The stocks must trade on any of the three primary exchanges, e.g., I omit OTC:BB and pink-sheet stocks
The choice of Friday evaluation is nothing specific -- it simply aligns with the weekend which is when I have more time to devote to analysis.  Of recent importance here is the gradual introduction of a $-volume criteria; I'm forward running a screening criteria that drops any stock with less than $700Ksh value, determined weekly at the close of Friday's market.  This will gradually be increased and performance of the database in terms of price action as well as the overall long-cash ratio (LCR, more on this later) behavior.  Fundamentally, I'd like to see the $-volume criteria move upwards north of $5Msh, nearly 7x where it is now, but I'll let the performance of the database dictate the ultimate cutoff.  The introduction of this criteria is simply to weed-out thinly-traded issues and ultimately, I envision it replacing the 100,000-share cutoff requirement.

Fundamentally, there are two systems within the GGT universe that I pay attention to:
  1. The Long-Cash Ratio system, which looks at the overall database in terms of the number of stocks that have a "Long" rating (meaning that it is okay to be holding these stocks), compared to the number of stocks with a "Cash" rating (meaning that if you are presently holding these stocks you're losing money).
  2. The Pricing System, which looks at the overall database in terms of aggregate stock value (closing prices of the individual securities).
The instantaneous value of these two systems is largely irrelevant; what is more relevant is the change that each of these systems is presently experiencing.  For example, knowing that a car is traveling at 60 MPH doesn't tell us about the anticipated road conditions in the near future.  If it's accelerating through 60 MPH, we have a reasonable expectation that the road is dry and that there are no curves ahead; if it is constant at 60 MPH we have a reasonable expectation that the road is straight but we may have some hills/valleys, or if  it's decelerating through 60 MPH, we have a reasonable expectation that the road conditions are changing, perhaps due to rain, curves, or simply being in uncharted territory.

I view the LCR and Pricing systems from this perspective. So, let's start there, specifically a view of history to date.

The LCR System

In the GGT universe, stocks get classified into two primary buckets:  
  1. Long, which simply indicates that the stock is outperforming the past with respect to some moving averages, and 
  2. Cash, which means that it is underperforming the past with respect to the same moving averages.  
How I derive these classifications is a topic for a different time, so for now, simply push the "I believe" button and know that we have over 2600 stocks that are either in a LONG status or a CASH status.

When I sum the stocks in each bucket, then take the ratio of Long:Cash, I have the raw LCR value.  Here is a plot of the LCR value since I started publishing the data back on September 11, 2008:



With all my graphs, right-click on the image to open in a new window or tab.

I've also superimposed the GGT Price index, which behaves very much like a broad index (its behavior is very close to the ETF Vanguard Total Index, or VTI).  I do this so that you can see the correlation between the LCR and the gains/losses of the price index.

I've artificially colored the LCR region below 1.0 in light red -- this doesn't mean that we should avoid trading in this area, it is simply to draw attention to the fact that more stocks within the database have a CASH ranking when the LCR is in this area.  If you look closely, you'll see that many a good move have started from within this zone, so establishing some rule that said you cannot invest when the LCR < 1 would be a terrible mistake.

What is important to note is that the instantaneous view of this snapshot shows that we just entered  the pink zone from above.  The Friday-close value of the LCR was 0.751, and indicates that 1083 stocks are LONG and 1443 stocks are recommended in CASH.  YOUR EYE catches the fact that we are falling into the pink zone, but what does this mean?

This next graph smooths the LCR using a 11-day exponential moving average:




Compare the chart above to the first chart in this blog entry -- you'll see that this simple 11-day smoothing of the raw LCR value gives us an easier view of visually correlating the performance of the GGT database to the GGT LCR.


What is important to me is that when the 11d LCR value starts and upward move, the database price typically moves upward too.  When the 11d LCR value reverses, the database price typically moves horizontally or downward.  It is the start of these upward moves which signify opportunities to invest on the long side, and presently, as you can see with the falling 11d EMA of the LCR, right now is not a good time to move into any long positions.

Where we "start" the next LCR run-up from in terms of real value of the LCR is important too.  Look closely at the figure above -- when we start a run from the pink area, we typically advance for 6-8 weeks before any pullback.  Yes, sometimes this fails, but generally, it has worked well as a confirming point.  Note that when we start a rally from above 1.0 we generally only last 3-6 weeks at best, and sometimes much shorter than this.

The most recent rally, which started on 11/30 or 12/1 (depending upon perspective), started from a 9d EMA value of the LCR of 0.999.  The waveform of the LCR performance after this date is not one of health, has certainly not occurred in the history of GGT, and does not confirm the continually-climbing prices.

The concept of units on LCR comes up from time to time -- people wonder what the units of LCR should be.  Recall that the LCR is a ratio -- the number of stocks that are LONG to the number of stocks that are in CASH.  A LCR value of 0.751 is the same as writing 1083:1443, or by stating that for every stock that has a long rating, we have 1/0.751 = 1.3324 stocks have a CASH rating.  The units are simply "# stocks long/# stocks cash".

This concept of a falling EMA plot opens up a discussion of the slopes of these LCR EMAs.  "Falling (or rising) indicates "change".  Recall that slope is "rise over run".  In our case, the "run" is easy -- it is 1 day in length.  Rise is a measure of whatever we are quantifying, and in this case, the LCR, so the units are "change in (# stocks long/# stocks cash) per day".  Hence, a change number of +0.15 represents that we have 15 out of every 100 stocks converting from a CASH basis to a LONG basis on that day (e.g., the database is expanding or becoming more bullish because the value is positive).  Conversely, a value of -0.07 represents that 7 out of 100 stocks are converting from a LONG basis to a CASH basis on a given day (e.g., the database is contracting or becoming more bearish because the value is negative).

I track the slopes of various-length EMAs of the LCR, so let's plot one of those, the slope of the 65d EMA of the LCR:



I've highlighted three key areas that we should remember concerning this present bull leg:  the September 2010 signal, the December signal, and the come-back-from-Xmas-vacation signal.  As you can see, the behavior of the LCR has been different for each of these events.  For the first one in September, the slope of the 65d EMA of the LCR moved positive on 9/2 and didn't look back until after 10/19, which is when it turned negative.  You should remember that it was relatively easy to make money in almost anything during this period.  As you can see, the GGT price index rocketed upward between 9/2 and 10/19, and the slope of the intermediate-term 65d EMA told us that we were "ok", e.g., it remained positive during this time.

[I'm not addressing the short burst on 11/4 that lasted only until 11/10, as this was a failed rally and it was pretty clear at the time that the stars were not aligned.  We got in, and we immediately got out ... ]

The signal on 12/3 worked until the week of the 14th, when volume started drying up.  Recall that a LONG rating requires both price and volume; without volume, stocks can slip into CASH status, and they are held there until volume in the market returns.  Despite this, the GGT price index continued higher, essentially unsupported by the LCR.

Finally, on 1/3/11, we had a major rally with price and volume, causing the highest LCR change since mid-December.  Volume had been high in this time frame, but note that the LCR had been struggling to maintain a positive level, again showing that prices are going up but that the underlying database of stocks was contracting, e.g., that it was getting smaller in the ratio of LONGS to CASH.

As we all know, this past week saw a major dive downward in terms of prices, and the LCR also reflected this, with a drop in the 65d slope of the LCR EMA.  Life isn't healthy for our portfolios at the present moment, because it's impossible to know how far we are going to drop, or even whether we will continue to drop.  Suffice to say, I'll continue to let the slope of the 65d LCR EMA guide me, and if it moves positive while the GGT price is moving positive, I'll have confidence to move back into the market.

This final figure will be my last commentary on the GGT LCR System:




Here, we see the table view of the previous slope of the 65d LCR EMA figure, but now I've expanded it to include the primary EMAs that I watch, those from the 5d to the 65d.  Compare the last three days of the 65d column (right-most) above to the last 3 days in the previous figure, and you'll see that the table above shows "red" or "bearish" and the previous figure shows the slope of the 65d LCR value in the pink zone (less than 0).  Hence, the table view above gives us some idea about the performance of the LCR EMAs, as a whole, in terms of change.  When all the slopes are pointing upward (bullish), we have a green condition, and when all the slopes are pointing downward (bearish), we have a red condition.

Look closely at the table above, specifically at the 12/1/10 signal.  We saw a crack in the ice around 11/29 for the slope of the 5d LCR EMA, then the thawing began on 12/1 and was confirmed across the board by 12/3.  THIS gave us confidence to move back long into the market.

Sometimes, the table view is easier to interpret.  Sometimes, the previous figure view is easier because of quantification.  It's subjective on what view I use, but the conclusions have to be equivalent because they show the same data...

All the LCR slopes are presently pointing downward, and have been for the last 3 days.  This is a short-term cautionary signal and I will not buy stocks long while this is the case.  When the database is contracting in terms of the LCR, the number of stocks in an uptrend that is supported by volume is getting fewer and fewer, implying that our stock-picking ability has to improve during the contraction.  I've not met one person who can consistently pick winning stocks on the long side during a market contraction.  If you are able to do this, then please introduce yourself to me by sending me an email to (pduncan [at] vt dot edu).


If/when a new long signal does develop, I'll need to see participation at the LCR level that clearly shows that we have an expanding database in terms of the LCR ratio.

==================

The Pricing System

I generally start with something taught to me by Ken Phillips, one of our NoVA group members and a long-term investor who has far more than he can teach me than what I could possibly impart to him.  One of the first questions about going long (or going short for that matter) is whether we are in an uptrend or a down trend.

The first question you should ask is "On what timeframe?"  If the answer is intermediate-termed lengths, then we are considering 6-week plus moves in the market before we have to make changes to our portfolio.  One way I keep track of the general market is by a simple plot of the 5d EMA and 65d EMA on a price series.  This works for indexes, like GGT, or for individual equities.

When you calculate the 5d and 65d EMAs on an index, you can establish the following rules:

  1. We are LONG on the market that comprises that index IF the 5d EMA > 65d EMA.
  2. We are CASH on the market that comprises that index IF the 5d EMA < 65d EMA.
It really is that simple.  

When we apply this rule to the GGT world, we get the following figure:


The figure above shows the status of when the 5d price EMA > 65d price EMA, as well as when this condition is not true.  We had a bullish confirmation in April 2009, exactly 1 month (22 trading days) after the market bottom, so you can see that although other systems may have signaled earlier, we had a confirmed rally and this was certainly the "all in" signal.

The spike downward occured on 11/4/09 and lasted exactly 1 day.  This is an issue with ALL EMA crossing methods -- they may whipsaw, but you need to stay the course and mechanically perform as the system indicates.  Looking back at my notes, we were just starting a new uptrend in prices but the LCR was falling, so this would have confirmed the dropping LCR and it appears that I moved to the sidelines for a few days.

The nail in the coffin to the major run of 2009 occurred on 5/19/2010, when we confirmed that we had a bone fide bear in the making.  We already had moved a major portion of monies to the sidelines at the end of April, so risk at this time was not very high.

As we all recall, this summer was a tough period, and due to events in my personal life, I took most of the summer off of trading and simply sat on the sidelines.

The next major bull signal was confirmed on 9/10/10.  Again, this was the "all-in signal", and while I was playing catchup (I was on a mountain on 9/1 when the LCR 65d EMA slope moved positive), but I jumped in  with both feet and was rewarded until I stepped out in mid-October.

As of the close of markets on 1/21/11, we are still in a confirmed bull market that confirmed on 9/10/10.  The difficulties that we are presently experiencing may result in a short-term pullback, or we literally could continue higher on Monday and never look back.  Certainly, keep an eye on my blog, as I will report on the status of 5/65 crossing if we get close to crossing from above.

Just like we have the ability to watch the slopes of various LCR EMAs, we can do the same with the GGT Price index EMAs.  The following table is exactly like the LCR EMA table and the interpretation is exactly the same, except we are measuring the amount that the database price rises and falls:



The table view above shows us that we presently have some issues with the present EMAs, as we have the start of a breakdown where the slopes of the shorter EMAs are already below 0.  The implications of this are easy -- if you have bought stocks within the last 5 or 8 days, because these EMAs are sloping down on the major GGT index, there is a greater likelihood that you are underwater on those positions.  Of course, actual mileage may vary, but if your stocks behave as the general market, then this is probably the case.

If the presentation above reverses, e.g., we start seeing green in the 5d and 8d positions, then we could be at a good buying point, everything else that I've presented not withstanding.  We need alignment on all signals in order for my confidence to come back, and I do think that we need to pull back and "reset" before we see any sustained bull.

My approach this week will be is selling into strength, as well as monitoring the Contra ETFs for surgical opportunities.  There are too many indicators telling us that we should be cautious.

=========

Remember, you are responsible for your actions, and I am not.  Please do your diligence and take ownership for your actions.

Regards,

pgd