Keep a channel-breakout to two lookbacks and test neighbor stability

How stance is defined

A channel-breakout changes stance when price penetrates the highest high or lowest low of a predefined lookback-window. That penetration is treated as the definition of a trend change.

The same idea can be specified as a stop-and-reverse rule that is always long or short and flips when the opposite channel is broken, as a long-only rule, or as a mixed stance that stays flat when neither an upside nor a downside penetration is defined.

Two lookbacks as the only critical knobs

Only two lookbacks, one for long entry and one for long exit, are treated as the critical knobs. Adding separate short-entry and short-cover lengths, or extra magnitude and time tests, is described as increasing overoptimization risk.

The worked long-only specification buys one point above the highest high of the buy lookback and exits one point below the lowest low of the sell lookback. A three-week high or low on weekly bars is used as a starting definition of a meaningful intermediate move. The weekly bar itself is treated as a noise filter, and the same logic is said to apply on other time frames.

What unsmoothed price accepts

Using unsmoothed price avoids indicator lag and supplies a market-defined stop. The same choice is expected to allow more noise, many false-breakouts, and crowded stops at obvious relative highs and lows.

Reading the optimization-table

A full one-to-four-week iteration of both lookbacks produces sixteen neighboring pairs. That optimization-table is used to check that the concept does not collapse when either window is nudged, not to select the single best cell.

Shorter lookbacks increase how often the rule fires. A one-week pair generated 95 trades versus 27 for a four-week pair, so the busiest cell is not treated as the preferred parameter-set.

The same windows on harder series

The same unadjusted windows are then applied to additional series chosen for choppiness and for a severe decline. The view in the historical workflow is that a trend-following procedure must be able to go flat in a past collapse and should not be judged only on persistently rising names.

Editorial. TradersWeek uses robustness to mean the same rule and the same windows remaining coherent across neighboring settings, assets, and difficult periods.

A filter treated as a second hypothesis

A longer-window moving-average-filter is introduced as a separate, simple confirmation idea rather than a fitted exception. It changed results favorably on one of the three test series and not the other two, so it is left unadopted pending further checks.