Custom Search

A nifty refinement of the straight line approach was developed by John Ehlers for his cyclic analysis work. He'd run a regression line (line A in Figure 4) through the last 20 days of data and record the R2. Then he would extend the line one day further back in time and recalculate the R2. If it stayed the same or increased, then the added day was consistent with the previous 20-day trend and he would go back another day to repeat the calculation. Stepping backward in this fashion (line B in Figure 4), he'd eventually find a point where the R2 decreased (the gap in Figure 4), an indication that the latest added price was inconsistent with the most recent trend.

All these straight-line methods are explications of our intuitive sense of expressing trend as direction, even if only horizontal.

The beauty of this approach is that it is indifferent to prices rising, falling or staying level. It points out to us that, quantitatively, trend can be persistent horizontal price levels (that is, lack of price movement) as well as moving up and down. Since this view of price action is more comprehensive than the intuitive "up or down," I prefer it. It subsumes the two-state model that prices are either moving or not moving and suggests a quantitative approach to defining "movement," abnormal departures away from the regression line or sharp changes in the slope and/or shortening of the period of the regression line.

Refinements, such as having a threshold level of change in the R2, using the slope of the regression line to define breakout, or even checking beyond the point of declining R2 for prices that would return the regression's coefficient to its previous values, are easy to imagine. It also ingeniously solves the time issue: how long must a trend persist to be a trend? Ehlers's approach has its limitations, but it's robust and rewards elaboration.

OTHER STRAIGHT-LINERS

In the vein of time, I also classify the various wave approaches (Elliott, Dow) as variants of the straight line approach because they subsume price movement into lines between peaks and valleys. I've always thought the best encapsulation of these approaches was Art Merrill's Filtered Waves , Basic Theory, which takes the straightforward approach that defining the percentage retracement would define the waves — the trends — for you (Figure 5). Longtime STOCKS & COMMODITIES readers may be familiar with these retracement charts from Art's monthly column, which usually uses them for comparison against various indicators. Software for generating them is included in RTR's Technifilter Plus or MetaStock Pro 2.5.

A 5% filter, for instance, would "filter out" all movements of less than 5% from the previous high or low. Here, it's easy to see that one's "scale" — and, effectively, one's time horizon (given normal movements) — can be set by the size of movement one seeks. I personally look for a 6% wave in the Standard & Poor's, so I'm unlikely to see a minor wave (a "minor" trend) of 1% or 2%. I'm also trading in a different realm than those following the Dow theory (10% to 25%).

BEYOND STRAIGHT LINES

All these straight-line methods are explications of our intuitive sense of expressing trend as direction, even if only horizontal. Once secure with line-drawing, averages are a refinement, since they reliably follow prices (Figure 6) without our intervention or judgment. In trends, they aren't quite straight but are close enough to make the connection for most people.

True refinements are French curves and fitted curves such as Bezier and regressions to exponential curves (a la Tom Kimball of Florida, of newsletter fame). Here, regression could also be used, perhaps searching through an entire family of potential curves for consistent fit over a given period of time. Again, price action is reduced to a simple line, but it's meant to be more indicative of the market's action. To wrap up, we usually identify trend after it's started by the slope of the lines we draw on our charts, whether straight or curved. However, if trend is persistence in price movement over time, trading range activity is a trend of sorts. We'll need a sharper definition if we just want price activity that's going somewhere. Our inability to isolate what we want is our own limitation: a predilection for simple lines.