Detecting hidden signals by Clifford J. Sherry, Ph.D.
Do you have a signal, such as a particular pattern of price movements, that is buried in noise like
random fluctuations or seasonal trends? If you believe this signal is time-locked to some internal event,
like a particular pattern of price changes, or an external event, like the beginning of a trading week or
month, you may be able to use a relatively powerful technique called averaging to detect your signal.
Averaging is done with a special-purpose computer and is used in such disciplines as neurophysiology or
electrical engineering to detect a signal buried in noise.
You can use this technique on raw prices, price changes, or categories of prices or price changes. For the
purpose of illustration, I will use categories of price changes as the data. The technique used to establish
the categories is described in the April 1986 issue of Stocks & Commodities ("Detecting a Dependent