V. 9:2 (76-76): LETTERS TO S&C

LETTERS TO S&C

GETTING A FIX ON LOOKBACKS

Editor,

I would like to comment on the excellent article by E. Michael Poulos entitled "Of trends and random walks" (STOCKS & COMMODITIES, February 1991). Mr. Poulos presents an extremely useful alternative to indicators calculated over fixed periods of time. I found this alternative very helpful in analyzing market indicators that I have begun to study and was able to apply his variable-length lookback concept to these indicators. The variable lookback techniques by itself provides a new perspective on developing accurate predictive models.

However, there is a second very important perspective presented in this article that is not really discussed. Mr. Poulos creates two indicators out of his data: a random walk indicator of highs and a random walk indicator of lows. This is the reverse of what is usually done; usually, two pieces of data (for example, advances and declines) are merged into one indicator (for example, an advance/decline line).

Markets can be modeled (despite some simplification) as two-control systems, with the two controls representing buyers and sellers. Many technical indicators lose this information as data are combined, resulting in one number. I would like to highlight the fact that Mr. Poulos has not done this in his model, and his indicators more accurately represent market movements as related to two control factors.

My appreciation for publishing this excellent article.

VICTOR KRYNICKI

Bahama, NC