Trend Strength and ADX: Separating Direction from Intensity

The average directional index (ADX) is the strength component of J. Welles Wilder Jr.’s Directional Movement System. It converts movement in successive highs and lows into positive and negative directional indicators, then measures the distance between those indicators. ADX is intentionally directionless. A high or rising reading can accompany either an advance or a decline, so direction must come from price structure or the directional indicators—not from ADX alone.

Derive positive and negative movement#

For each period, compare the current and previous extremes:

Up move = Current high - Previous high

Down move = Previous low - Current low

Positive directional movement (+DM) equals the up move only when it is positive and greater than the down move; otherwise it is zero. Negative directional movement (-DM) equals the down move only when it is positive and greater than the up move; otherwise it is zero. If the moves tie, both are normally zero.

This rule prevents one period from contributing to both directions. True range is calculated separately as the largest of high-low, absolute high-previous close, and absolute low-previous close.

Smooth movement into DI lines#

The conventional period is 14. Initial true range, +DM, and -DM are summed across the first 14 observations. Wilder’s recursive smoothing for each sum is:

New smoothed value = Prior smoothed value - Prior smoothed value / n + Current value

Directional indicators normalize movement by true range:

+DI = 100 × Smoothed +DM / Smoothed true range

-DI = 100 × Smoothed -DM / Smoothed true range

When +DI exceeds -DI, positive directional movement dominates the smoothed sample; the reverse holds when -DI is higher. A crossover changes that ordering, but can occur repeatedly in a range.

Move from DX to ADX#

Directional index (DX) measures the absolute separation between the DI lines:

DX = 100 × absolute value of (+DI - -DI) / (+DI + -DI)

When the denominator is zero, software needs an explicit zero-handling convention. Initial ADX is usually the average of the first (n) DX values. Subsequent values use Wilder smoothing:

ADX today = ((Prior ADX × (n - 1)) + DX today) / n

ADX ranges from 0 to 100. Readings near or above 25 are often used to describe meaningful trend strength, while readings below 20 are often associated with weak direction. Those are conventions, not universal boundaries. Instrument, timeframe, and regime can shift their usefulness.

Worked example: strength without direction#

Suppose 14-period Wilder-smoothed totals are:

Smoothed true range = 28.0

Smoothed +DM = 8.4

Smoothed -DM = 2.8

Then:

+DI = 100 × 8.4 / 28.0 = 30

-DI = 100 × 2.8 / 28.0 = 10

DX is:

100 × |30 - 10| / (30 + 10) = 50

If the prior ADX is 24, the updated 14-period ADX is:

((24 × 13) + 50) / 14 = 25.86

Directional movement is positive in this snapshot because +DI is above -DI, and trend-strength measurement has risen above 25.86. If the DI values were reversed—10 and 30—DX would still be 50 and the new ADX would be identical. That symmetry demonstrates why ADX cannot identify direction.

Practical ADX checklist#

1. Confirm Wilder’s DM, true-range, smoothing, and zero-division conventions.
2. Allow enough history for DM smoothing, initial DX values, and initial ADX.
3. Read +DI versus -DI for directional dominance and ADX for strength.
4. Compare current ADX with the instrument’s own historical range.
5. Distinguish rising ADX, falling ADX, and absolute ADX level.
6. Evaluate DI crossovers separately in trends and ranges.
7. Include delayed signals, next-bar execution, spread, slippage, and gaps in tests.

A threshold should remain fixed during an evaluation period instead of moving to explain each historical outcome.

Limitations and false signals#

ADX is highly processed: highs, lows, and closes create DM and true range; those values are smoothed; their ratios create DI and DX; DX is smoothed again. The result can lag a new trend and remain elevated after momentum has peaked. A falling ADX means directional separation is weakening, not that price must reverse.

DI crossovers can whipsaw in overlapping markets. An isolated gap or limit move can raise directional movement and true range in ways that persist through smoothing. Different session definitions, adjusted data, futures rolls, and insufficient warm-up history can produce mismatched readings. Threshold optimization is another risk: selecting 21 for one instrument and 27 for another after observing outcomes may capture noise. ADX also omits volume, liquidity, valuation, and scheduled-event information.

Key takeaways#

• ADX is the directionless strength component of the Directional Movement System.
• +DI and -DI describe directional dominance; their separation creates DX.
• Reversing the DI lines leaves DX and ADX unchanged.
• Rising ADX indicates strengthening directional separation, not necessarily rising price.
• Multiple smoothing stages create lag and require substantial warm-up history.

This guide is general education, not personal investment advice or a recommendation. Trend-strength signals can fail or arrive late, and gaps, liquidity changes, slippage, and leverage can turn false signals into substantial losses.