Signal Type	Trade Date	Index	Return since issued
		Nasdaq 100
		Russell 2000
		S&P 500

Market Update

The major averages moved higher for the third week in a row, sending the Dow Jones Industrial Average to a new all-time high and the S&P 500 to its highest level since early 2001. The market showed its underlying strength by closing higher Monday despite news that North Korea had just tested a nuclear bomb. Not so long ago, such news would have sent stocks reeling, as was the case on July 5 when the Nasdaq Composite lost 1.7% after North Korea tested ballistic missiles. There is no question that the investing climate has improved considerably since then, as the major indexes have been on an uninterrupted uptrend since late July. Markets only paused briefly Wednesday after Alcoa kicked off the earnings season by releasing a disappointing third-quarter report. The weakness did not last, however. Buoyed by oil prices that hit their lowest level of the year and a Federal Reserve's beige book report showing continued economic strength, stocks shot higher Thursday, with the Russell 2000 gaining more than 2% on the day. Building on their momentum, the main indexes posted additional gains Friday, capping another strong week for stocks.

First initiated by large-cap stocks, the 3-month rally has now broadened to smaller companies, as is illustrated by the fact that the Russell 2000 closed 3.09% higher on the week. For their part, the Nasdaq 100 and S&P 500 respectively gained 2.51% and 1.19%. All three indexes still rest well above both their respective 50-day and 200-day exponential moving averages (EMAs). Our current Buy signal remains active.

Trend Timing School

The correlation coefficient

A distinctive characteristic of Trend Timing is that the trend we are looking for is that of the broad stock market. We believe that, as evidenced by historical data, broad stock market indexes move in tandem most of the time. We have long distinguished between which direction a trend is going from how strong the movement is. We have several tools to gage the amplitude of stock market moves, with volatility measures such as standard deviation (see "Volatility creates opportunity" weekly update sent on 9/29/2006) and, of course, the strength indicator that is the basis of our World Index Ranking.

With respect to market direction, we have the TimingCube signal as a proxy for mid-term trends, but if one wants to know how closely a particular index moves in comparison to the broad markets represented by the signal, one needs to delve into correlation. Correlation tells us the mutual relation between two indexes, or how "synchronized" they are. For the most part people look at charts to determine the correlation of various investments or indexes. In these pages we have presented numerous graphic illustration of good and bad correlation, such as "Correlation of Nasdaq Composite with other indices" or "The Trend is contagious".

Many of our subscribers find that interpreting a chart is more an art form than a science, and they would much prefer to have an analytical tool in addition. Here comes the Correlation Coefficient, or CC for short. Statisticians never cease to amaze us... The correlation coefficient calculates how closely two data sets move together. CC is a number which varies between -1 and +1, with +1 being perfectly correlated. For a daily correlation coefficient, "perfectly correlated" means that the two data sets move in the same direction by the same amount every day during the observation period. For example, two investments gaining 0.1% every day will have a CC of 1. Conversely, an investment which does exactly the opposite of another is said to be inversely correlated and earns a CC of -1. Zero is reserved for data sets which do not have any particular relationship, be it direction wise or amplitude wise.

Table 1 below lists the correlation coefficients of major world markets of periods of ten years, five years, and 2006 year-to-date, as measured against the Nasdaq Composite index. There are many interesting observations to be made from the historical data. Most notable is that correlation between world markets has been increasing over time, as a sign of an increasingly interconnected planet. Yet, differences abound, from almost perfectly correlated (Nasdaq 100) to almost inversely correlated (China). Also see today's related FAQ of the Week "Why are Chinese and Russian markets not in the World Index Ranking?".

One surprise is how low the correlation exhibited by the Dow Jones Industrial is of late.

Table 1: The evolution of correlation between world stock markets
	From: To:	Last 10 years 01/02/1997 10/06/2006	Last 5 years 01/02/2001 10/06/2006	2006 Year-to-date 01/02/2006 10/06/2006
Index Symbol	Description	Daily Correlation Coefficient
^IXIC	Nasdaq Composite	1.00	1.00	1.00
^NDX	NASDAQ 100	0.99	0.93	0.98
^MID	S&P 400	0.18	0.76	0.88
^RUT	Russell 2000	0.28	0.79	0.83
^N225	Japan	0.47	0.78	0.81
^ATX	Austria	-0.02	0.61	0.80
^KS11	South Korea	N/A	0.55	0.76
^DWC	Dow Jones Wilshire 5000	N/A	0.94	0.74
^OMXSPI	Sweden	N/A	0.86	0.72
^GSPC	S&P 500	0.87	0.95	0.61
^MIBTEL	Italy	N/A	0.84	0.60
^GDAXI	Germany	0.83	0.76	0.58
^BVSP	Brazil	0.19	0.72	0.58
^STI	Singapore	0.44	0.78	0.58
^FCHI	France	0.84	0.76	0.51
^BFX	Belgium	0.44	0.76	0.48
^GSPTSE	Canada	N/A	0.78	0.48
^FTSE	UK	0.75	0.79	0.47
^SSMI	Switzerland	0.63	0.78	0.46
^TWII	Taiwan	N/A	0.70	0.45
^DJI	Dow Jones Industrials	0.64	0.92	0.21
^BSESN	India	N/A	0.65	0.21
^AORD	Australia	0.15	0.67	0.19
^KLSE	Malaysia	0.13	0.57	0.16
^IBEX	Spain	0.74	0.84	0.14
^MXX	Mexico	0.09	0.61	0.08
RTS.RS	Russia	N/A	N/A	-0.12
^HSI	Hong Kong	0.62	0.88	-0.17
399300.SZ	China	N/A	N/A	-0.65

The reason correlation is so important with our form of investing when applied internationally is that it helps us stay away from markets that behave different or inverse from world markets, and the TimingCube signal. Throughout history local or regional events and conditions have caused certain markets to fall out of synchronicity with world markets. A perfect example of this occurred during the 1996-1998 East Asian financial crisis, the period covered in Table 2 below. The previously high flying tiger nations of the "Asian economic miracle", such as Malaysia and Singapore were hard hit when international investors lost confidence in their currencies and equities and bailed out in droves. While most Western economies were in the midst of a decade long bull market, most Asian countries declined severely, resulting in strong inverse correlation with other world markets. Malaysia had a correlation coefficient of -0.80, almost perfect inverse correlation.

Table 2: Inverse correlation of markets during East Asian financial crisis
	From: To:	11/11/1996 12/31/1998
Index Symbol	Description	Daily Correlation Coefficient
^IXIC	Nasdaq Composite	1.00
^GSPC	S&P 500	0.97
^DJI	Dow Jones Industrials	0.95
^MID	S&P 400	0.95
^NDX	NASDAQ 100	0.93
^FTSE	UK	0.92
^IBEX	Spain	0.91
^SSMI	Switzerland	0.89
^FCHI	France	0.87
^BFX	Belgium	0.87
^GDAXI	Germany	0.87
^AORD	Australia	0.79
^RUT	Russell 2000	0.68
^ATX	Austria	0.45
^MXX	Mexico	0.40
^BVSP	Brazil	0.16
^TWII	Taiwan	-0.30
^KS11	South Korea	-0.31
^BSESN	India	-0.34
^HSI	Hong Kong	-0.54
^N225	Japan	-0.67
^STI	Singapore	-0.73
^KLSE	Malaysia	-0.80

The trouble with the correlation coefficient is that it is tainted by measuring both the direction of the movements but also the amplitude of those movements. For example, two indexes that move up or down together 100% of the time but with one moving up or down much more than the other one would be poorly correlated, i.e. a low CC. Also, because of strength differences, the highest correlated index is not necessarily the best performer.

Since correlation numbers can be misleading, it is always a good idea to check using the "Performance with individual security or index" tool on the "Results" page. The general rule of thumb is that if the performance using the TimingCube signal with the Long and Short strategy is higher than that for buy and hold, you can conclude that during the observed period there is a positive correlation. At one extreme you have China with a CC of -0.65, and when you plug the index ticker symbol (399300.sz) in the ticker field you see that trading with a TimingCube Long and Short strategy would have lowered the buy and hold results (41.50% versus 53.96%). You don't want that, even if 41.50% lucked out to be a great result. Markets with low correlation numbers can still work out well in the context of our mid-term signals. Take Mexico , with a CC of +0.08 would have returned 46.01% with signals versus 26.42% for buy and hold. Of course, the indexes benefiting the most are the highly correlated ones. The Nasdaq 100 is a virtual twin of the Composite with a CC of 0.98, and this helps the long and short traded result to reach an impressive 20.88% when buy and hold over the same period only generated 3.88%.

The conclusion of all this is that the correlation coefficient is an imperfect measure which needs to be looked at only in combination with indicators of strength and volatility.

FAQ of the Week

Question: Why are Chinese and Russian markets not in the World Index Ranking?

There are many stock markets around the world but for the purposes of the World Index Ranking we narrowed the field with some of our requirements:

• Availability of a minimum 5 years of reliable and publicly accessible index data
• A stock market generally correlated with major world markets
• The existence of investment vehicles which approximate the index

We hear requests for many additional countries but, probably because of their recent performance, we hear about China and Russia the most. It turns out that they both have a number of strikes against them, starting with unavailability of sufficient index historical data.

Since we do not have long term data the only conclusions about correlation we can reach are strictly short term. And they are not good. Chart 1 below depicts the Chinese and Russian markets versus the Nasdaq Composite over the last 12 months. Their performance has been vastly superior to North American markets, their volatility is much higher, but the correlation is lacking. This is further confirmed by two of the lowest and negative correlation coefficients of all the indexes listed in Table 1 above, -0.12 for Russia and -0.65 for China.

Chart 1: Chinese and Russian markets versus Nasdaq Composite

Index Yahoo!Finance Symbol China Shanghai-Shenzhen 300 399300.SZ Russian RTSI RTS.RS Nasdaq Composite ^IXIC

Lastly, the only available ETFs for these countries ( for China; and for Russia) are of the closed-end ilk, and they do not track their country indexes very well.

We will certainly keep looking at these countries and others with a view to include them when they meet our minimum requirements.

Warm wishes and until next week.

The TimingCube Staff