1986issue C021-11
Cycle-aligned directional trend indicator
Classic directional movement averages daily up and down high-low differences twice over a span treated as a typical half-cycle, then unsigned the result into Average Directional Index. This construction keeps a signed interim directional ratio and uses two quarter-cycle averages so Directional Trend Indicator is meant to crest and trough with price when a dominant cycle is present.
- Differencing successive highs or lows leads a cyclic series by about a quarter turn. Pairing that phase lead from differencing with averaging is how the construction tries to put directional movement back in step with price.
- A quarter-cycle average of the up-move and down-move series, then a quarter-cycle average of the signed interim directional ratio, is intended to leave Directional Trend Indicator with a net half-cycle lag and in phase with price.
- The stated rule marks a downward hypothesis after a crest and an upward hypothesis after a valley, and only when the absolute value of the indicator exceeds 0.7.
- A sine-like Directional Trend Indicator is treated as evidence that a usable dominant cycle is present. Splitting one half-cycle average into two quarter-cycle stages is presented as a way to keep turning marks usable when the cycle-length guess is wrong by about half.
From an unsigned index to a signed indicator
Classic directional movement averages daily up and down high-low differences twice over a 14-day span, treated as a typical half-cycle. It then forms an unsigned index as the absolute MU minus MD gap divided by their sum, and averages that index again to obtain Average Directional Index.
This construction replaces the unsigned index with a signed interim directional ratio, IN equals (MU minus MD) divided by (MU plus MD). That ratio is then averaged to obtain Directional Trend Indicator, a signed directional series built from averaged up and down high-low differences, then averaged again, so the finished oscillator is meant to crest and trough with price when a cycle is present.
Spending the phase lead
On a pure sine wave, a half-cycle moving average lags by about 90 degrees, while a full-cycle average is zero and recovers an underlying trendline. Differencing successive highs or successive lows leads a cyclic series by about 90 degrees. That phase lead from differencing, paired with averaging, is the mechanism used to restore phase alignment with price.
Averaging the up-move and down-move series over a quarter-cycle average, then averaging the interim directional ratio over another quarter-cycle average, is intended to give Directional Trend Indicator a net half-cycle lag and place it in phase with price. The dominant cycle is the prevailing periodic length used to size those averaging windows.
A two-stage average and a decision rule
The stated Directional Trend Indicator decision rule marks a downward hypothesis after a crest and an upward hypothesis after a valley, and only when the absolute value of the indicator exceeds 0.7.
Splitting one half-cycle average into two quarter-cycle stages is presented as a way to reduce sensitivity to a wrong cycle-length guess. On an 80-unit synthetic sine wave, turning marks stayed usable when the cycle input was 80, 40, or 120. The derivation treats that as tolerance to about a 50 percent cycle-length error.
A sine-like shape as the cleanliness check
On synthetic sawtooth and reduced-harmonic waveforms, a sine-like Directional Trend Indicator shape is treated as evidence that a usable dominant cycle is present and that turning marks remain well placed at cycle inputs of 80 and 120.
The construction is described as most reliable when other spectral components sit about 10 dB, or about one-third the amplitude, below the dominant cycle, or when Directional Trend Indicator itself looks like a sine wave. The dominant cycle is also the length used to judge whether other rhythms are weak enough for the construction to stay interpretable.
Average Directional Index in this construction
In this construction, Average Directional Index remains a second long average of unsigned directional movement. It collapses toward a late trendline of the indicator rather than marking turns in phase with price.
All readings on this track · 56 readings
- 1986Cycle-aligned directional trend indicator
- 1987What crossover and directional entry rules actually compare
- 1988A directional-line cross needs a trend filter, an extreme-point rule, and a dollar stop
- 1988Constructing true range by offset addressing
- 1988Constructing directional movement from bar range
- 1988Average directional index construction: recursive smoothing and lookback offset
- 1988Staged Average Directional Index construction with Relative Strength Index confirmation and stop alerts
- 1988Average Directional Index construction with frozen true range and directional rules
- 1991Constructing the average directional index from range expansion and true range
- 1991Constructing five-session forecasts from stochastic, ADX, and MACD inputs
- 1993Constructing the average directional index from directional movement and true range
- 1993Confirming n-bar breakouts with ADX and DX filters
- 1994Constructing a Bollinger band-width trend filter
- 1994A pre-trade checklist that can refuse a long three ways
- 1997An ADX threshold and a moving average as a trend filter
- 1998Regime filters for mutated indicators
- 1999Building the average directional index from range extension and true range
- 2000Evaluating ADX, RSI, and moving averages in a multi-stock warehouse
- 2000Stochastic pop as a filtered continuation setup
- 2000Onset and exit from one average directional index
- 2002Joint ADX and MACD readout for trend strength and direction
- 2003Adaptive Donchian breakout with implied volatility and volume
- 2004The average directional index as a regime gate for the relative strength index and the stochastic oscillator
- 2004Constructing true-range-specified volume as a directional filter
- 2005Constructing a multi-filter penny stock breakout procedure
- 2005Construct one playbook that flips with session regime
- 2005Combining Bollinger Bands, the average directional index, and Fibonacci retracement on currency pairs
- 2006Assembling an adaptive price zone from double-smoothed averages
- 2006An ADX strength gate for MACD and the stochastic oscillator
- 2007Directional movement as a filter plus trigger
- 2007Constructing a veto-first trend permission stack
- 2007ADX gates for trend end, range, and reversal
- 2008Constructing a nine-cell directional-ratio grid
- 2008Average directional index and directional trend indicator lookbacks as trend-filter parameters
- 2008A holding-matched market lens from averages and directional-line crosses
- 2008A nine-cell directional scoreboard for multi-horizon entries
- 2010Building a Vortex Indicator from high-low distances
- 2010Constructing ADX, RSI, and MACD price filters
- 2011Constructing a volume zone oscillator with a moving-average and Average Directional Index regime filter
- 2011A volume zone oscillator conditioned by an Average Directional Index filter
- 2011Candlestick names need volume-price, ADX, and moving-average checks
- 2012Clustered average-directional-index traces as a trend-start filter
- 2012Average Directional Index cluster filters for trend-start signals
- 2012Confirming a trend start or turn with a triple ADX cluster
- 2013Constructing a late-entry stack from a signed DMI oscillator
- 2013A directional oscillator and its stochastic as a stacked timing filter
- 2013ADX cluster lookbacks are a locked specification, not a chart label
- 2013Combining moving averages, stochastics, and ADX in a daily scan
- 2015Assembling the Average Directional Index from directional movement
- 2016How an Average Directional Index filter and a breakout entry form one procedure
- 2016Score RSI and stochastic crossings only when ADX confirms the trend
- 2018Constructing an ADX filter for intraday breakouts
- 2018An ADX volatility gate for prior-day breakouts
- 2019Exponential deviation bands with a moving average, RSI and ADX
- 2020A normalized-slope trend filter from linear regression
- 2020Gating volatility-momentum divergences with a Trend filter