Monday, May 28, 2012

Update for Tuesday, May 29th -- Short Term Signal Moves Long

Effective with the close of markets on Friday, May 29th, GGT models are indicating that a short-term long entry into the market is warranted.  This is based on two key indicators:

1) the 4d average of the Long-Cash Ratio has been crossed from below by the daily LCR, and
2) the slopes of the 2d, 3d, and 5d exponential moving averages of the LCR are now positive.

When we get this type of signal we typically can enter on the long-side as close to or below the closing price on the day of the signal.

Overall, statistics for this signal are okay at best but not the best of the GGT systems:

There are a fair number of trades with this methodology -- 66 -- and there are favorable independent metrics such as the System Quality Number (SQN), the Mathematical Expectation (ME), and Pessimistic Return Ratio (PRR).  We want these values to be above 1.7, 0, and 1.0 respectively, and as you can see, PRR is on the threshhold.  Nevertheless, all show a positive edge.

Note that if using the GGT index (close proxy is the ETF Vanguard Total Index, symbol VTI), the worse signal of the 33 losses is a realized -7% and the best trade of 33 wins is + 12.5%.  Median is 0% so the tails are balanced -- this is truly a balanced system.

The evolution of the SQN tells me how the system has been behaving recently:

The number of trades is listed across the bottom and the SQN is on the y-axis.  The system seems to bounce between 2.00 and 2.65, which is higher than the recommended level of 1.7 and indicates a better-than-chance edge at long-term success with this model.

Working against me is that the intermediate-termed timer as well as the long-termed timer are both quite negative and are indicating CASH:

This means that we are swimming up stream with this signal, and when trout swim up stream, they tend to get eaten by the bears.

In terms of familiar price and slope charts, here you go:

Note that the 5d EMA slope of the LCR is now positive, and note that it is on an upward trajectory.  Any failure of this below 0 will cancel our short-term buy signal.

Somewhat disconcerting to me is that we have a significant narrowing of the price slopes for all time frames. We can break up or down from here -- up is good/bullish, down is not good.  Note how these have fallen back below the zero line or are right at thresholds -- it's never good to buy at thresholds because of potential whipsaws.

Time will tell.

I intend to enter a 1% @ risk position in VTI near the open on Tuesday morning, with a stop loss at -7% below by entry point.



Monday, May 14, 2012

Update for Monday, May 14th

All my models are indicating CASH (not SHORT) at the present time.

The GGT Price Index fell a significant amount on Friday (-2.6%) on volume that was -2% below the 50d MA of volume. This is a huge move for the index and places the daily value under the 200d MA for the first time since 1/18/12. This action will also cause the 200d slope to turn negative, which is an incredibly bad sign for the bulls.

As with all my images, right click in the picture to open in a new tab or window.

The GGT Database Strength indicator, which rolls up price change, volume, and rate of price change, dropped a bit more than 1%. This shows that while prices did drop, volume was not present in the drop (average volume only), and also that the rate of change of the drop was not significant across the board. Friday's action was a controlled distribution of stocks and I interpret as a simple unloading of positions going into the weekend.

You can see from the table above that the slopes of the various price EMAs are all negative, and have been since 5/4. You can also see that on the right side of the table that the slopes of the slopes (acceleration of prices) are back and forth -- we do not have panic developing, but we do have a steady trend down.

There is nothing in the price slope model that indicates a floor is in place or that we are about to bounce.


The Long-Cash Ratio (LCR) slope model continues to indicate a contraction of the database, e.g., we should not be buying stocks on the long side at the present time. You can see this in the slopes of the LCR all remaining negative:

Of interest to me though is that we are starting to see some positive indications that the contraction is stabilizing. The right side of the table, which is the slope of the slopes and represents the acceleration of stocks moving to LONG recommendations or CASH recommendations, shows that we've had two days of positive-moving acceleration -- and this is a pre-requisite of the database starting to move back upwards in price. This is the primary indicator which tells me that we may be abating in the present drop, as two days of upward acceleration is becoming significant for a short-term bounce.

Futures are down as I write so this will obviously get tested and possibly negated today.


All my timers are indicating that we should be in CASH:

The shortest time that the 13d / 65d timer has been negative in the past is 19 consecutive days and we just completed day 2. While we could whipsaw from here and move positive, my intermediate-termed Elder timer is hugely negative and has no chance of going positive unless we get a +500 point day in the markets -- probably not today.


As many of you know, I've been watching divergence of prices with EV for Pascal's universe. He publishes the list of equities daily in his DIVA.xls file, which is available for subscribers at, and if you are not looking at this file, you're missing a learning experience.

Positive divergence is a rising price and outflow of volume -- large institutions are leaving the equity but it not yet reflected in price.

Negative divergence is the opposite -- large buyers are entering and price has remained low or is dropping.

When I net the two groups against each other a balance for the day is formed and it looks at the average behavior over the last 3 days. Hence, if we have a greater number of negative divergent equities, we are net buying in the market and visa versa, or so my theory goes.

When I apply different lengths of correlation with my GGT index to the net value of divergence, I get an interesting plot:

Normally, a rising price in GGT is heavily correlated with a falling value in net divergence -- sellers step in and begin exiting as prices move higher. The converse is true too -- buyers step in as prices drop. This gives us a negative correlation value, and it typically lies in the -0.4 to -0.6 range on a 34d lookback.

But take a look at what happens if we simply consider just an 8-day lookback: we have periods where the correlation moves strongly positive, indicating that there are points where the market prices AND effective volume align. Also note that I've circled some of these peaks, and they typically correspond to turning points in the GGT price index. Not 1:1 exactly, but they do alert us that things are happening.

On Friday, we jumped into positive territory for this 8-day correlation. This means that because the GGT price index fell -2.6%, and over the past 8 days has been falling a great deal, that we are seeing a significant OUTFLOW of EV that is in line with GGT prices. 

My crystal ball is as good as yours but I think we're at a short-term capitulation point with Friday's and possibly today's action (I'm writing before the open on Monday, 5/14), and I would expect to see the markets move up here shortly.

We'll see.

Make it a great week!



Saturday, May 5, 2012

Update for Monday, May 7th -- Transition to Cash

With the close of markets on Friday, May 4th, all GGT models are transitioning from LONG to CASH. None of my indicators are suggesting the short side of stocks, although a number of -3x leveraged ETFs did signal "New Long".

I will close my QQQ position at the open Monday for a loss of about -3.1%. This position was derived from a test model that I am developing and I was exposed at a level of 0.5% equity with a -6.61% stop loss. While above the SL level, the macro model signal is kicking me out.

I will continue to hold my UVXY positions and will add to those on any sign of market strength.

For those of you who are watching my TSP model over my shoulder, it too has signaled a move to cash in the C-, S-, and I-Funds. Ensure you read the details here.


My primary model is based on a combination of the Elder Force Index Timer and 13/65d moving average cross timers which I have developed. Let's look at the individual timers:

Yesterday (Thursday close), there was a mismatch between the 13d Force Index calculated with an exponential moving average (faster) and a simple moving average (slower). Today (Friday close) these two are in agreement, and with the 13d and 34d GGT price slopes negative (red), we are confirmed CASH in this model.

As with all my images, right-click on the figure to open in a new tab or window.

The graph above shows the state of the subtracting the 13d EMA from the 65d EMA. The scale on the left is the one to pay attention to -- if we drop below 0, then the long-term trend I track will also be DOWN. We are close to this transition and any sustained weakness in the markets will be reflected here quite quickly.

The combination of these two timers results in a signal which requires the 13d and 65d crossing to be LONG and the Elder timer to be LONG for entry. If either of these transition to CASH, we're out. The Elder timer has confirmed this latter state, hence, we're out.

 I've listed the most recent statistics on the chart above -- we did fall a bit because this last cycle is a net loser on the broader index, but overall, performance is quite sound. Recall that we want:
  • PRR > 1.0 and preferably above 2.0
  • ME > 0 and preferably above 1.0
  • T-Test > 1.7 and the larger the better
  • # trades > 20-30 and we're working on that
  • Compounded Rate of Return as large as possible

Here's a snapshot of the detailed metrics to date - make your own decisions:

 Unfortunately, this last signal has resulted in a loss of -1.85% not including commissions or slippage, but this is well within the worse recorded trade level of -2.41% and is a bit larger than the average losing trade level of -1.58%. 

I am compelled to take the trades indicated by the model as long as the metrics hold. So far, so good, despite this last loss.

Finally, in terms of equity and all the timers I track, this combo timer represents a good trade off between performance and number of trades, and I have confidence in the model:

 In the end, the short-term timer is in CASH, the intermediate-timer is in CASH, and only the long-term timer is holding long:


The GGT price index fell -1.72% on volume that was +7% above the 50d MA of volume. This is significant. Coupled with Thursday, where we saw a -1.27% decline on volume that was also +7% above the 50d MA, we have a one-two punch that tells us to exit the markets. Hence, the models are responding.

There is nothing bullish in the GGT Price Slope Models:


The LCR Slope Model is also quite bearish:


Monday should be relatively easy. There are no compelling signals aside from sensitive 3x leveraged instruments telling us to go short. Hence, cash is king, and aside from my UVXY positions, I intend to focus on other things.

I will be in Spokane, Washington all week next week. Postings and Dropbox files will be updated as time allows. 



Friday, May 4, 2012

Update for Friday, May 4th - Short Term Timer = Cash, Intermediate Timer = Mixed (Warning)

We've had a state change in my timer system:

The short-term 4d Long-Cash Ratio (LCR) timer has moved to cash. This occurs when the database begins to contract, and the daily LCR value falls below a 4d simple moving average (SMA). 

The intermediate-termed Elder Force Index timer has moved to a mixed state. A mixed state occurs when their is disagreement between the 13d Force Index calculation using a faster exponential moving average (EMA) and a slower simple moving average. The faster EMA value is now negative, while the slower SMA method is still (barely) positive. This is a significant warning.

The longer-termed 13d / 65d price index crossing is still positive. 

Correspondingly, we are in a long-term uptrend, but the intermediate time frame is under tremendous pressure and the short-term is telling us to exit if we took the most recent signal.

The natural question is "what to do?" 

To answer this, I draw upon past model performance:

As with all my images on this blog, right-click to open in a new tab or window.

The Combined Long-Term (13d/65d) and Elder Timer model is still long, and is shown at the far right. We are a bit underwater from our starting point (-0.13%, no commissions or slippage), but in terms of practical, sustained performance, this model is telling us to hold the line and remain long. The worse realized loss from this model is only -2.41% since 11/25/08, and while intraday drops will certainly be below this value, and while we have yet to see the next worse trade (a model's worse trade has not yet occurred), this model is telling us to sit pat. The confidence in this model is high -- the t-Test/SQN value is 2.65, and we want values above 1.7 -- this is telling us that we are far better than chance in performance. Further, the mathematical expectation (ME) value is well above 0 at 2.62, which is very, very good and tells us that this model has a solid edge to the long side. Finally, the Pessimistic Return Ratio (PRR) is 2.77, well above 1.0, and again, confirms the model's long-term efficacy.

The Short-Term and Re-Entry Timer models both had conditions which were triggered with the 4/25 buy signal. They are now both indicating that we should exit, so it is not surprising that their gains for this last trade are equal. You can see from the statistics that the 4d, short-term model has the poorest performance overall and the weakest PRR value, so despite the good ME value, we can rank the signals from the 4d model as the lowest of the 3. On the other hands, the Re-Entry Timer has performed quite well at getting us in and out of trades if the master signal is LONG, and correspondingly, this one is telling us to exit.

So we have a mixed situation -- the combined model is long, the short-term model is indicating cash but is not a solid performer, and the re-entry timer, which performs well, is telling us to exit. It's a coin toss but the coin is biased to 2 against 1, and I intend to reduce 1/2 my positions at the open today. I will keep a nominal position on the table due to the combined timer.

I'm also holding a QQQ position which is based on a new test system. I've held the position since 4/17, and am now slightly underwater. The test system is indicating to hold QQQ on Friday, which I will do. I will not adjust the position size, as this runs independent of the systems described above. Hence, you may see me refer to a QQQ position, and that is the rationale of continuing to hold.

In my TSP account, which I discuss here, there are no signal changes going into the day on Friday. This is a long-term portfolio and I can only make two transitions per month, so I need to protect those moves.

Make it a great weekend!