Range trading

A family of methods built on the observation that a market which has repeatedly turned back inside a band has, historically, tended to keep doing so until it does not. The conventions are consistent across the literature: the band's boundaries are read from prior turning points, positions are established near an edge rather than in the middle, protective levels are placed beyond the boundary, and the opposite edge is used as the reference for an exit. It is the direct counterpart of a breakout method, which is built on the opposite expectation about the same bars.

Because the band is drawn by eye, most range methods add a second, mechanical filter, and the choices differ from author to author. An oscillator reading near an extreme, a narrowing average true range, a trend strength reading below a conventionally cited threshold and a count of how many times each boundary has held are all used. None of them defines a range. They are conventions for deciding when one is present, and they disagree with each other on the same chart often enough that the choice of filter changes the count of trades a method would have taken.

Two difficulties are structural rather than avoidable. Every range ends, and the bars that end it look identical to the bars that formed it until after the fact, which is why methods of this kind conventionally place a protective level beyond the boundary rather than at it. And the cost side compounds: a method that takes a position at each edge pays the spread and any commission on every round turn, so the same band produces more chargeable events than a method that holds through it. Practitioners disagree about whether a range can be identified while it is forming at all, since the evidence for one is a sequence of failures to leave it.