« Trimming, Rather Paring, Apple Again :-) | Main | Disney's Movie Studio Suddenly Looking Good »

December 02, 2005

December Model Signals

There were no changes to the signals from Northlakes Market Cap and Style models for December. The signals remain mid cap and value, respectively. Therefore, clients continue to own the S&P 400 (MDY), and the Russell 1000 Value (IWD) and the Russell 2000 Value (IWN). The market cap model has been drifting very slightly toward large caps over the last few months so the ratio of IWD to IWN is 3 to 1...

The indicators in the market cap model are split with six favoring large cap and four favoring small cap. The trend indicators favor small caps due to strength throughout most of 2005 in small and mid cap indices. Since the trend indicators are overweighted in the model, essentially we got a split decision. When the model offers a mixed signal it defaults to mid cap. Looking a little deeper, I categorize the individual indicators into three buckets: stock market, economic, and interest rate. Even at this level, the signal remains mixed with the stock market indicators slightly favoring large caps, the economic indicators slightly favoring small caps, and the interest rate indicators split right down the middle.

In the style model, there was a moderate shift toward growth this month but the signal remains value. Out of nine indicators, six currently favor value but the trend indicators and the advisory service sentiment indicator both shifted from a value signal to a growth signal for December. The trend indicators are picking up outperformance for growth indices since May. The sentiment indicator flipped to growth when bearish sentiment, which had gotten quite high prior to the current rally, bounced off its lows. Overly negative sentiment is a sign that a rally may be at hand. Growth generally leads sentiment-based rallies, probably due to the higher beta in growth sectors like information and communications technology.

Over the last few months, Northlake's models have been mixed in their accuracy. Looking back at monthly data going back to the early 1980s, it is clear that the models are most accurate when the signals are strong. With the weak and fluctuating signals generates since summer, I am not all that surprised that accuracy has been mixed. Nevertheless, I am frustrated. Here is some data:

The market cap model has been flashing mid cap for the last three months. Overall, this has been a decent call offering marginal upside relative to Northlake's S&P 500 benchmark. Last month, MDY was up 5.7%, ahead of the 4.4% return for the S&P 500 as measured by the S&P 500 ETF (SPY). Small caps as measured by the Russell 2000 ETF (IWM) rose 5.8%. The good October performance for MDY allowed it to pull slightly ahead of SPY for the three month period with IWM bringing up the rear.

The style model has just been wrong over the past two months during which it has been calling for value over growth. I monitor style performance looking at the Russell 3000 Growth and Value and Russell 1000 and 2000 Growth and Value. During November, regardless of the measure, growth comfortably outperformed value by 1-2%. October also had an inaccurate call in favor of value so over the two month period the outperformance for growth expands to 1.25-2.75%. Seasonality favors growth off the October lows based on market history but I never outguess the models and thankfully the rising market has provided nice returns even a little something has been left on the table.

Posted by Steve Birenberg at December 2, 2005 01:41 PM in Models

Comments

Thanks for the information on the models- over a long period of time these models seem to have been beating the market. Do you know if there been extended periods where they underperformed in the past, or flashed the wrong signal like the Style model has the last 2 months? I would not be particularly concerned that these models have lost their shine until we got to 6-8 months of being wrong. Of course, everyone has their pain points. I am curious, for those Blackjack players out there, you know how you increase your bet when the cards are clearly in your favor? Is there a corollary here in terms of the 0 to 100 signal on Steve's models? Just a thought , executing that tactic could be tricky, but thought I'd postulate.

Posted by: Paul Mokdessi at December 4, 2005 09:51 PM
Post a comment









Remember personal info?






Schwab Login