Pairing same-horizon oscillators with a walk filter

A shared horizon first

Most technical readings in the study depended on changing cyclic content. A method aligned with the current cycle produced timely marks. A method that was not aligned produced out-of-phase marks.

Editorial reading: compare the two oscillators only after they share one sampling interval and lookback. The relative-strength index is a bounded momentum reading from ordered closes over a fixed lookback, used as a same-horizon overbought and oversold forecast. The stochastic oscillator is a range-position oscillator from recent highs and lows over a defined sampling interval, used as the baseline overbought and oversold forecast.

Walk as a companion filter

The walk construction compared an up series and a down series with a random-walk baseline. A reading above 1 on either series was treated as a trend in that direction. That pair is the random-walk-index: up and down displacement series scaled by path length and average true range so each reading can be compared with a random-walk baseline and with other price series.

The up-walk and the down-walk are the two sides of that index. The walk denominator used path length and average true range so gold, currency, and equity series could be compared on one scale.

Standalone permutations of the walk were not judged a robust system. The useful output was overbought and oversold information placed beside a standard oscillator.

When extra oscillators repeat the same area

When a stochastic oscillator, band-position, walk index, channel index, and percent-R were set to the same timeframe, they marked the same overbought and oversold areas. Band-position here is percent-b: the close located between an upper and a lower statistical band, used as a second overbought and oversold scale.

Editorial reading: on one shared horizon, a relative-strength index that only restates those same extremes is redundant. The pair is then a repeated forecast, not new information.

Lookback-information at two horizons

Lookback-information is the extra or missing turning-point content that appears when the same method is recomputed at a short horizon and a long horizon. At an 80-bar lookback the walk lost some turning-point content but gained four items, while band-position lost content and gained none. Both still carried trend-direction information.

Same-period walk and band-position were described as nearly identical for trend identification. The extra walk content looked closer to cycle-free in the 20-bar versus 80-bar comparison.

From agreed extremes to one procedure

Mean-reversion is a testable rule set that treats clustered overbought or oversold states as conditions for fade, exit, or abstention rather than as a standalone trend engine. Agreement of the walk and band-position on an overbought or oversold state was treated as a clear mean-reversion condition. Many strong moves in the examples began from those readings.

When the two studies diverged, walk slope was used for probable direction. Fading a rising walk trendline was treated as a weaker tactic than using pullbacks to enter or add.

An hourly diary combined walk-band extremes, a stochastic-style oscillator panel, and price structure into one entry, exit, and skip procedure. The diary included cases where only the walk was oversold.