Framework for Ts Model
November 17, 2008

Starting today we are going to re-introduce the Ts (Trend sector) model and portfolio on a daily basis. Each day we calculate two numbers, the ten year history of which enables us to gain a valuable insight into predicting stock market behavior. For example, on Friday the Ts indicators closed at 0.02 and 0.33. The first number records the total number of stocks that are in definable Up trends to those that are in Down trends. The closing number of 0.02 means that there are presently 50x (1.00/.02) as many stocks in Down trends as there are in Up trends. Over the past ten years this number has varied from a low of .01 on Oct 27th 2008 to a high of 7.95 on June 5th 2003. Typically a number above 1.00 is commensurate with an Up trending market. The second number is a broader definition of the technical strength of the overall stock market. Over the past ten years this number has varied from a low of 0.23 on Oct 10th 2008 to a high of 1.82 on January 13th 2000. Again, typically a number above 1.00 is commensurate with an Up trending market. We feel that unlike discussing a particular level on the Dow or S&P 500, the combination of these two Ts indicators offers us a more powerful insight to the “internal” condition of the stock market. While the study of these two indicators and there correlation to stock market behavior offers us many observations, the overwhelming conclusion that we come away with is that good things tend to happen when both indicators are above 1.00 and that the opposite is true when both indicators are below 1.00. For example, if we look back over the past year, we can see that both indicators initially moved below 1.00 on Nov 2nd 2007 which marked the beginning of the downturn. We also note that the only time that the indicators got back above 1.00 was April 3rd of this year and that this new Up trend only lasted until June 9th. If all we had done over the past year is be long the market during that brief two month period and short the rest of the time, then we would be very profitable today. If we were instead to place our fundamental hats on over the past year, we would have noted that before the bear market began, stocks had already suffered through four years of p/e multiple contraction and that relative to Treasury yields, earnings yields were as high as any time in history. We would also note that monetary policy was favorable, that insiders were buying heavily and that investor sentiment and mutual fund outflows were already at levels indicative of market bottoms not tops. And yet here we are in the midst of one of the worst bear markets in history. The lesson here is that fundamental analysis alone is not sufficient and that there must be in place a rigid set of rules that keep us from adding exposure in down trending markets. So is it simple enough to buy when both indicators are above 1.00 and short below 1.00? While this would have certainly worked very well over the past year, it is important to appreciate that most technical indicators are “lagging” indicators, and that the great majority of trend following systems or moving average crossovers etc, fail because they are either whipsawed to death, fail to have a profit trigger or never take a counter trend position. While buying above 1.00 and shorting below 1.00 is not without merit, it is not in and of itself sufficient to earn profitable returns on a constant basis. The following is a brief synopsis of the framework of our Ts model and the critical observations we utilize to predict market behavior. Not only do we look at just the two Ts indicators but also where the underlying index is in relationship to the 10 week/50 week crossover. We have noticed that Up trends will be more sustainable if the 10 week is above the 50 week average and similarly for Downtrends when the 10 week is below the 50 week. A broad portfolio of varying sectors would rely on the underlying position of the S&P 500, but as we progress, the goal is to update Ts portfolios on a variety of sectors.

1) Uptrend - when both indicators are above 1.00 - this is a time when many stocks are either “breaking out” or in continuous Up trends. We have found this is a great time to buy positive relative strength stocks on pullbacks. The 10 week average has to be above the 50 week average on the underlying index before long side exposure can be increased significantly in the Up trend condition.

2) Downtrend - when both indicators are below 1.00 – often this will occur after a pullback, and while the initial reaction maybe to buy as stocks are suddenly cheaper, we have found that just like Nov 2nd 2007 and June 9th 2008, the prudent thing is to do the opposite. Admittedly about half the Down trend signals are false, but it is the ones that aren’t that can inflict such lasting damage and must be avoided. This is the time to momentum short the weaker relative strength names as they are breaking down. We have found that only once the second indicator has fallen below 0.60 should we begin to throttle back on the level of short side exposure. If the 10 week has fallen below the 50 week average of the underlying index, we have found that the Downtrend condition can be particularly severe. (just look at the 50% decline in the Morgan Stanley Commodity Index (CRX ) since the 10 week broke the 50 week on August 29th) and it is only during these times when short side exposure should be increased significantly.

3) Counter Trend - we have found that often when the second indicator reaches 1.60 or above the market is ripe for a pullback. However as this pullback is often short in nature, and often in the confines of a much longer sustainable Up trend, we have had difficulty incorporating into the model any major counter trend move on such a signal.

Whenever the first signal falls to 0.25 or lower and then moves back up to 0.70 or higher, we have seen a great tendency for this “reversal” action to take hold over the next calendar month and for stocks to continue higher. Whenever the second indicator falls below 0.45 it is time to start making counter trend buys, and certainly when we fall below 0.35 we have almost always seen a strong up move over the following calendar month. When stocks first move out of very oversold conditions, one of the most striking facts is that the lower relative strength, more value oriented stocks actually outperform the stronger relative strength stocks, at least for that initial blast off move. Some of the most powerful counter trend rallies occur when the 10 week is below the 50 week, so we will buy an Oversold condition irrespective of where we are in the 10 week/50 week crossover position.

If you simply followed the framework of the above Ts model we could show quite clearly how an investor would have outperformed the S&P 500 many fold over the past ten years. This does not mean it always works, and indeed there are many times when one can be made to look quite foolish, but over the long run it works very well and at times outstandingly well. It is important to recognize that this is an evolving process and that there is no such thing as a perfect system. What we are attempting to do is provide solid time tested indicators that keep us in when the going is good, get us out when problems surface, and attempt to pick up the pieces when there is a good chance the intermediate carnage is over.

We have been working on a variety of stock screens that computerize the different characteristics wanted in the individual stock picks based on the prevailing market conditions, as determined by the Ts indicators. Again this is a continuing process and while we have found some great and consistently profitable stock selection screens, we will always be trying to improve results. For example, we recently ran a screen of all technology stocks. We bought ten stocks that had a relative strength index above one but had lost the most on a percentage basis over the previous three days. We then simultaneously sold ten technology stocks that had a relative strength index of less than one but had gained the most on a percentage basis over the previous three days. We held each stock for five trading sessions. From October 1st 1998 to the end of 2003, the long screen gained an average of 1.67% per week, while the short screen lost 0.61% per week. That’s an astounding 2.28% per week profit for over four years on a neutral position. Since 2003, however, we have found that while the gains are still decent, the long screen has outperformed the short screen by only 0.76% per week. We will be constantly back testing stock screens to refine the process and offer more insights to stock behavior. Typically, however, we are looking to buy pullbacks in Uptrends and sell rallies in Downtrends. The only exception is in the deeply oversold condition, where it has historically paid to buy washed out stocks.

Each day we will now be writing an analysis of where our Ts indicators stand, what the past results tell us and what we hope to expect moving forward. We will be starting a new Ts portfolio completely based on these indicators and that will be run on a “real time” basis. Over time we hope to introduce “sector” portfolios that utilize the individual 10/50 week crossovers of the underlying index as well as our Ts indicators. For example, Oil Service is the OSX, Retail is the RLX, Commodity is the CRX, Technology is the MSH etc.

Moving forward we will abandon the “real time” trading portfolio and concentrate solely on our Ts model. This model is not designed to find individual stock market winners, but to successfully manage a portfolio of stocks by increasing or decreasing long or short side exposure based on the prevailing market condition as designated by our Ts indicators.