« June Model Signals | Main | Crash Winding Down Very Successful Run For Lions Gate »

June 09, 2005

Ned Davis Research On CNBC

This morning, the #2 strategist at Ned Davis Research (NDR), Tim Hayes, discussed NDR's latest views on the stock market. A Northlake client astutely noted that Tim was recommending large cap value while Northlake's models just switched from large cap value to large cap growth. Given that Northlake's models are produced by and based on NDR analysis, the client wanted to know why there was a conflict. The answer lies in the disciplined implementation of the models for Northlake clients....

...NDR's own models did in fact recommend a shift from value to growth for June. In an email I received today, NDR notes that the shift from value to growth in the main Style model did take place but in their “opinion” it is a weak reading favoring growth and might reverse given the underlying trends in some of the individual indicators. On CNBC, Tim Hayes was merely expressing this opinion even though it was at odds with his own models.

The whole point of using models is to enforce discipline and EXCLUDE “opinion” from the decision-making process. Opinion is subject to the emotions created daily in the stock market as prices move around. Given the tendency of human emotions to overreact to what is happening now, I think emotions and opinion work against investors. I do not try to overanalyze and anticipate what may happen next. I back-tested the models without second guessing or imposing my "opinion" on the monthly signals and those backtests show superior performance. Now that I have been using the models to manage money for more than 18 months, I am confident this is the right approach and that the backtests work in the real world.

Whenever the models are at transition points, moving from one signal to another, there will be some volatility in the signals. The likelihood of increased portfolio turnover is higher as month-to-month model signals may switch more often. This adds modest trading costs but over the long-term, the average holding period for the models is around six months so turnover and commissions are not an obstacle to performance.

The bottom line is that I do not try to outguess or anticipate the next move in the models. I think that is the best approach and the backtests and real world experience supports that approach.

Posted by Steve Birenberg at June 9, 2005 02:10 PM in Models