by John Sweeney
Having recently gotten a handle on the natural order of prices (i.e., they are random, stationary and
dependent), the next order of business has been to explore dependence. Naturally, traders are interested in
those forms of dependence which give us an edge in trading. Several forms come immediately to mind:
cycles, trending, basing and congestion. Of these, MESA and EPOCH exploit the existence or
non-existence of short-term, cyclical components of a price series.
MESA, an earlier product, will search through your prices trying to identify cyclical elements with
frequencies of 20 to 55 days. Given that result, it will then estimate and plot a two-week price forecast.
MESA was well received, but needs regular usage before an analyst can become familiar with its
strengths and weaknesses. Its value could not really be nailed down because trading signals were always
given in hindsight, and changed as future prices changed the optimal cycle and its half-cycle average. To
boot, you could not tell MESA which cycle to use in its forecast. The result was the program's output was
more art than quantitative assessment.
EPOCH overcomes these weaknesses. It allows you to specify any time period (or epoch) and it will
evaluate any assumed cycle between 8 and 32 days in length. The program then creates a table showing
the number of trades, the profit and profit/trade for that length. The results may also be graphed for the
screen or printer, large or small.