Skip to main content

Trend Indicators (MACD, ADX, MA Systems)

Updated Jun 23, 2026 at 8:47pm

Research Draft Medium 1,141 words

Trend indicators are the family of technical tools that attempt to identify the direction and/or strength of a sustained price move and filter out the noise of short-term oscillation. The three most widely used are MACD (a momentum oscillator built from moving averages), ADX (a non-directional measure of trend strength), and moving-average systems (dual/triple MA filters). Their shared, defining property is that they are lagging — all are mathematically derived from price history (raw price, or moving averages of it), so they confirm a trend that is already underway rather than predict one. This creates their core tension: smoothing buys clarity at the cost of timeliness, and the same smoothing that filters out whipsaw in trends produces whipsaw of its own in sideways ranges.

How they're calculated / formed

MACD (Moving Average Convergence/Divergence). Developed by Gerald Appel in the late 1970s. Three components, default settings (12, 26, 9), all EMAs (StockCharts ChartSchool):

  • MACD line = 12-period EMA − 26-period EMA
  • Signal line = 9-period EMA of the MACD line
  • Histogram = MACD line − signal line

The MACD line is the spread between a fast and slow EMA, so it is itself a momentum measure of the underlying trend. The signal line is a moving average of that momentum, which makes it smoother and slower, and the histogram visualizes the gap between the two.

ADX (Average Directional Index). Created by J. Welles Wilder in New Concepts in Technical Trading Systems (1978); default period 14. Built in stages: directional movement (+DM/−DM) is derived by comparing consecutive highs and lows; +DM and −DM are smoothed (Wilder's smoothing) and divided by smoothed True Range, then ×100 to give +DI and −DI; DX is computed from the normalized difference between +DI and −DI; ADX is a smoothed average (EMA-like) of DX. The double smoothing means ADX lags heavily — StockCharts notes it can take on the order of ~150 periods of data before ADX values fully stabilize.

MA systems. Trend filters built from two or more moving averages — e.g. a dual-MA system (price/fast MA vs. slow MA) or triple-MA system. Signals come from crossovers and from price's position relative to the averages. The mechanics, period choices (SMA vs EMA, 50/200, etc.), and the golden-cross / death-cross conventions live in the Moving Averages node — see that node, not duplicated here.

How to read them

MACD — three canonical readings (StockCharts): 1. Signal-line crossover — bullish when MACD crosses above its signal line, bearish when it crosses below. The most common timing trigger. 2. Zero-line (centerline) crossover — MACD above zero means the 12-EMA is above the 26-EMA (uptrend bias); below zero is the reverse. Slower, higher-conviction. 3. Divergence — price makes a higher high while MACD makes a lower high (bearish), or price a lower low while MACD a higher low (bullish).

ADX — read strength only, never direction. The widely used convention (per StockCharts, and broadly attributed to Wilder's system): ADX > 25 indicates a trend is present; ADX < 20 indicates no/weak trend; 20–25 is a gray zone. Many analysts use 20 as the single key level. Direction comes separately from the +DI/−DI cross: +DI above −DI = bullish edge, −DI above +DI = bearish. A rising ADX means the prevailing trend (whichever direction +DI/−DI shows) is strengthening; a falling ADX means it is weakening, regardless of price direction.

How they're used in practice

The dominant professional use is confirmation and filtering, not standalone signal generation. ADX is the textbook regime filter: traders take trend-following signals (MA crossovers, breakouts) only when ADX is above its threshold, and switch to mean-reversion tactics when ADX is low — this directly addresses the "indicators whipsaw in ranges" failure mode by sidelining trend tools when no trend exists. MACD is used both as a momentum-confirmation overlay (is the trend's momentum still expanding? watch the histogram) and as a crossover trigger, usually combined with a directional filter so signals are only taken in the direction of the larger trend. MA systems supply the baseline trend direction that the other two confirm. The canonical combination is a direction tool + a strength filter + a momentum/timing tool — e.g. price above the 200-day (direction), ADX > 25 (strength), MACD signal-line cross (timing). Style-specific operational mechanics (exact entries, stops, hold periods, multi-day risk) belong to the Swing Trading branch — cross-link there.

Standing & evidence

These indicators are near-universally adopted, built into every charting platform, and ~50 years old (Appel, Wilder both late-1970s), so their standing as standard tooling is uncontested. What is genuinely contested is standalone predictive edge. Honest points an expert keeps straight:

  • The 25/20 ADX thresholds are a convention, not a law of markets. StockCharts presents them as such, and they are commonly attributed to Wilder, but Wilder's original book leaned on a directional-movement "rating" rather than a clean 25/20 cutoff. Treat the numbers as widely-used heuristics, not validated constants — markets and timeframes differ.
  • MACD divergence is unreliable as a timing tool. StockCharts explicitly cautions that bearish divergences are commonplace in strong uptrends and bullish divergences in strong downtrends — i.e. divergence fires repeatedly against a trend that keeps going. It flags possible exhaustion but routinely fires early or falsely.
  • All three are lagging by construction. They cannot lead price; the best they do is confirm faster than a human eye.

Strengths & limitations

Strengths. Objective and rule-codable; excellent at confirming an established trend and at filtering (ADX especially) so a system avoids fighting a range. Combining a direction + strength + momentum tool is more robust than any one alone.

Limitations. Lag means late entries and late exits — in fast reversals these tools give back the move's edges. The single most common misuse is treating a lagging confirmation tool as a leading predictive one — e.g. trading MACD crossovers or divergences mechanically in a choppy, low-ADX range, where they whipsaw relentlessly. Their edge is strongly regime-dependent: meaningful in trending conditions, weak-to-negative in ranges. Parameter settings (12/26/9, 14) are defaults, not optimized truths, and curve-fitting them invites overfitting.

System relevance

Within Delvantic, these are confirmation inputs, not verdicts. The Augustus trade-setup agent should consume them as evidence weights alongside the regime context — e.g. an ADX reading corroborating the regime engine's trend/range classification, MACD confirming momentum direction — never as a standalone trigger. The hard caveat to carry downstream: because all three lag and whipsaw in ranges, Augustus must condition them on regime (don't trust trend-confirmation tools when the regime is range-bound). MA-system mechanics cross-link to the Moving Averages node; swing-specific entry/stop/target mechanics defer to the Swing Trading branch.

Sources