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

Cumulative Returns since First TimingCube Live Signal () as of

Index	Long Only	Long Only with Margin	Long & Short	Long & Short with Margin	Buy & Hold
Nasdaq 100
Russell 2000
S&P 500

Market Update

It has been a punishing week for stock investors. Marred by escalating tension in the Middle-East, earnings warnings and record-high oil prices, the major averages lost ground all week, suffering some of their worst losses in months. With rising volume accompanying the plunge, there is little doubt that institutional investors are major participants in the broad sell-off. Disturbing economic news did nothing to alleviate investors' fears: retail sales slipped by an unexpected 0.1% in June and the University of Michigan consumer sentiment index fell, hinting that the economy may experience a significant slowdown in the second half.

For the week, the Nasdaq 100 and Russell 2000 respectively lost 4.67% and 3.96%. As for the S&P 500 , it finished 2.31% lower. All three indices now rest below both their 50-day and 200-day exponential moving averages (EMAs). We now have a Cash signal in effect. You can read more about it in the Trend Timing School article below.

Trend Timing School

Preservation of capital

As all of you are painfully aware, we are experiencing an unprecedented, at least since 1989, market phase during which our Model has struggled. While our longer-term trend following Model is not designed to catch short-term market starts and stops, it got tricked into detecting fake trend reversals causing, for the first time in our history, four signals in rapid succession of which three were losers. In case anyone missed it, we are repeating here the message which accompanied our Cash signal:

"For full disclosure, our Model actually issued a Sell signal on July 13, 2006 after the market close which we have decided to override with a Cash signal, which means that we are liquidating all our long positions and going to cash. In this unique market, our Model appears to detect overlapping up and down trends. These mutually exclusive conditions, other technical indicators and common sense dictate that we go to cash until a clear trend emerges."

As you can imagine, this new turn of events caused some consternation and certainly raised a lot of questions, some of which we will attempt to answer here for you. But before we go any further, we would like to thank all of you for your continued support through these tough times. We are touched by the many supportive comments we receive and impressed with how many of you remain confident and committed to the Trend Timing system. We also want to send a very big thank you to the many who have spontaneously sent us suggestions for enhancing the Model. All ideas are welcome and we are busy testing them (see Continuous improvement below).

Post mortem

So what did really happen over the last two months? We have of course been conducting our investigation to identify what went wrong and what we can do about it. We do not mean any of this as excuses or as spin of recent events. Rest assured that we continuously and relentlessly research and study the markets and our Model to improve it. Anything we learn from this painful period is and will be applied in the future.

In retrospect, here is how we see the last few weeks. The uptrend which started in October 2005 petered out in early May this year rewarding our Buy signal with decent gains, especially for those invested in the Russell 2000 (18.67% gain) or most international indices. The May 11, 2006 Sell signal was as close to perfect as any trend follower can wish for. We now know that the May 31 Buy signal was erroneous and in the process caused a whipsaw. One major component of our Model is volume as it evolves and interacts with prices. It appears that volume was distorted by options expirations. This tricked our model into viewing a more bullish environment than it really was and issue a Buy. This was clearly wrong and we now know why and can avoid it going forward. In this instance, it really was a problem with the input data, not the Model itself.

Then there is the June 29 Buy signal which arguably could not have been avoided. This was a very clear and unambiguous rally which most timers and media recognized as such. If we tweaked our Model to filter out this type of signal we would probably wipe out every other winning trade. What went wrong there is that, as it frequently does, the market did not oblige and instead of continuing on the confirmation, it reversed course. When the June 14 low from which the rally started was undercut on July 13, it was a clear indication that the rally was dead. However, what we feel went wrong is that our Model took too long to realize that the rally had fizzled.

The Cash signal

Our disclosure of details about the functioning (or non-functioning as some would put it) of our Model such as the fact that a Sell was triggered which we overrode, has caused some to write us off. We have always favored complete transparency and honesty. We know it is better to keep our subscribers informed, even if it is with bad news or the admission that we got whipsawed for losses, mistakenly once, and for losses that were higher than they should be.

The purists are crying foul because manually overriding a mechanical system is a capital sin. But we beg to differ, as TimingCube probably has one of the highest concentrations of purists anywhere. Contrary to appearances, our decision to override the Sell signal was not at all triggered by recent losses or because we listened to our gut and got scared out of the market.

In fact, what we did not know is that our Model was telling us that there was a problem. Simultaneously with the Sell signal, the conditions for a reversal with a new Buy signal were also present, meaning that we could easily get whipsawed again. Entering this Sell would have been as if "pre-loaded" for a whipsaw because the two trends are present concurrently, with the balance shifting slightly back and forth from day to day. In such an environment, going to cash is the reasonable thing to do until a clear direction emerges. We will also take the time to analyze how we can improve on our Model to avoid such problems going forward.

Yes, the Cash signal is an admission that in addition to Buy and Sell phases there is also a third state represented by this new type of Cash signal for periods of trendlessness or of conflicting trends. In situations where the technical conditions do not apply, the Cash signal may cause a Buy or a Sell to be temporarily delayed or ignored.

Continuous improvement

As mentioned above, there are many avenues of research for us to enhance our Model. We know that there is no silver bullet, there are multiple solutions for multiple issues, and they are not all going to be resolved tomorrow morning. Rather it is a continuous improvement process which did not begin yesterday and will continue in the future. Some enhancements have already been identified, tested and implemented, others we have not found yet. Will we ever be able to eliminate all anomalies and imponderables? Of course not, but we know that our Model is strengthening further because of this episode. There are many promising ideas and leads we are following right now.

Easy, "low hanging fruit" enhancements such as adjusting the input data for non-meaningful events was a tweak that was rapidly tested and put in practice. It is really nothing new since we already correct data for anomalies such as short trading sessions, we now are adjusting for other known factors such as option expiration days, and end of quarter because they do not represent what institutional investors are doing.

The trick is that in order to make a change in the Model, we have to make sure it does not degrade past performance. That is why such enhancements are hard to come by. Also, because our past performance is what it is, we would not change past results because of a change. We would only implement the change going forward.

Since in the stock market the only certainty is that it will continue to change and evolve, we know we will never be perfect but with our continuous improvement process our Model can adapt and become stronger over time. We are confident that the TimingCube model is still valid and will continue to produce superior results in the future.

FAQ of the Week

Question: What is an UltraShort ETF?

As promised last week, we duly inform you of the introduction of the four UltraShort ProShares ETFs on July 13, 2006. These are the first ETFs designed to provide magnified short exposure (-2x) to well-known market indices, and they represent a very practical alternative to "double inverse" index mutual funds. These funds complement the double and inverse funds announced earlier (see "What are ProShares ETFs?" published in the June 23, 2006 Weekly Update) and complete the initial set of 12 which ProShares received SEC approval for. The four funds are as follows:

• UltraShort QQQ ProShares (QID), double the inverse of the Nasdaq 100 Index
• UltraShort S&P500 ProShares (SDS), double the inverse of the S&P 500 Index
• UltraShort Dow30 ProShares (DXD), double the inverse of the Dow Jones Industrial Average
• UltraShort MidCap400 ProShares (MZZ), double the inverse of the S&P MidCap 400

On a side note, we have heard many questions about the absence of Russell 2000 tracking products in the ProShares family of ETFs, and whether we should wait for them. We would not recommend waiting, because the Russell 2000 ETFs may never come. For whatever reason, ProShares have decided to substitute the S&P MidCap 400 index for the Russell 2000. For all practical purposes the two indices have tracked each other very closely over the years and we believe the MidCap provides a good enough approximation of the Russell 2000 for our purposes.

Warm wishes and until next week.

The TimingCube Staff