by John Ehlers
Are there basic differences in the cyclic behavior of various commodities? Although theory predicts
short-term cycles will come and go, there is no theory predicting different cyclic "personalities" for
different commodities. My experience has been that cyclic trading is successful for some commodities
but not for others. My curiosity about the cyclic behavior of different commodities was aroused during
the development of a Maximum Entry Spectral Analysis program (MESA) that provided backtesting.
Using the program, I measured the cycles of 12 commodities over the past two years. I found that each
commodity does have its own cyclic personality. Moreover, commodities within the same category
(meats, grains, metals) do not necessarily have similar cyclic personalities.
I used perpetual daily contracts containing 500 records, and the average of the high and low for each day
determined the price from which the cycle was measured. The program then measured the short-term
cycles for each 50- day period, starting with record number 50. (The program needed the first 50 records
to make the first measurement.) The dominant cycle and cyclic content were measured daily for records
50-500, a total of 451 measurements.
"Cyclic content" is a signal-to-noise ratio over 6 decibels. Noise is all of the non-cyclic energy with
periods of less than 5 days and greater than 40 days. (Noise by this definition includes trendlines as well
as the random day-to-day fluctuations of price.) My experience is that a cyclic content of at least 6
decibels is useful for trading. At lower values, the noise greatly influences any predictions based on
cycles, reducing the accuracy of the predictions. Therefore, I discarded measurements having a cyclic
content of less than 6 decibels.