ADX Indicator Explained

At a glance

What it measures: Trend strength, regardless of direction
Range: 0 to 100, typically 10 to 50 in practice
Common thresholds: Below 20 means no trend; above 25 means trending
Direction comes from: The +DI and -DI lines, not ADX itself

Key takeaways

ADX rises in strong downtrends as well as strong uptrends, because it measures the strength of directional movement, not its sign.
Its most valuable use is as a regime filter: enabling trend strategies when ADX is high and reversion strategies when it is low.
ADX lags substantially because it is a smoothed average of smoothed values, so it confirms trends rather than anticipating them.
A rising ADX matters more than its absolute level, since the level depends on the instrument and the timeframe.
Combining ADX with a trend-following entry addresses that strategy family’s main weakness, which is whipsaws in flat markets.

How ADX is built

Directional movement each bar:
   +DM = current high - previous high   (if positive and larger)
   -DM = previous low - current low     (if positive and larger)
   Only the larger of the two is recorded; the other is zero.

Smooth over N periods and divide by ATR:
   +DI = 100 x smoothed(+DM) / ATR
   -DI = 100 x smoothed(-DM) / ATR

DX  = 100 x |+DI - -DI| / (+DI + -DI)
ADX = smoothed average of DX over N periods

Note the structure:
   +DI and -DI carry the direction
   DX measures how one-sided the movement is
   ADX smooths DX, which is why it lags

The three components and their relationship.

The double smoothing explains the indicator’s character. ADX is slow, it confirms rather than predicts, and its readings are stable enough to use as a regime classification rather than as a signal.

Reading ADX correctly

ADX level	Interpretation	Strategy implication
Below 15	No meaningful trend; tight range	Trend strategies will whipsaw; consider range or reversion
15 to 20	Weak or emerging trend	Marginal; most trend filters exclude this zone
20 to 25	Trend developing	Common threshold for enabling trend entries
25 to 40	Established trend	Favourable for trend following and breakouts
Above 40	Very strong trend	Trend continuation likely; reversion strategies dangerous
Above 50 and falling	Trend may be exhausting	Tighten trailing stops rather than reversing

Using ADX as a regime filter

The most valuable application is deciding which of two strategy families to enable. Trend strategies fail in flat markets; reversion strategies fail in trending ones. ADX provides an objective, mechanical way to classify the current regime.

Trend mode: ADX(14) above 25 and rising. Enable breakout and trend-following entries; disable range and reversion entries.
Range mode: ADX(14) below 20. Enable range and mean-reversion entries; disable breakout entries.
Transition zone: ADX between 20 and 25. Take no new entries from either family, or halve position sizes.
Direction: Supplied separately: price relative to a 100 or 200-period moving average, or +DI above -DI.
Recalibration: ADX levels differ by instrument and timeframe. Examine the historical distribution for your market and set thresholds at percentiles rather than at fixed numbers.

DI crossovers and their limits

The classic system buys when +DI crosses above -DI with ADX above a threshold, and reverses on the opposite cross. It is a legitimate trend-following rule with the usual properties.

It lags. Both DI lines are smoothed, so crossovers occur well after the move begins.
It whipsaws without the ADX condition. The DI lines cross frequently in flat markets, which is precisely what the ADX filter exists to prevent.
It is one of many equivalent trend rules. A DI crossover, a moving average crossover, and a Donchian breakout will agree most of the time. Using several is duplication.
It needs a separate exit. DI crossovers as exits are late. An ATR trailing stop generally performs better.
It requires diversification. Like all trend rules, it works across a portfolio of uncorrelated markets rather than on a single instrument.

Practical notes

The default 14-period setting is standard. Longer settings produce smoother, later readings; shorter ones are noisy enough to lose the indicator’s main advantage.
ADX levels are not comparable across instruments or timeframes. A reading of 30 may be typical for one market and exceptional for another.
Use percentile thresholds rather than fixed levels when applying ADX across a diverse universe.
A falling ADX from a high level does not mean reverse; it means the trend is decelerating, which is a reason to tighten stops rather than to take an opposite position.
ADX is a poor entry timer. Pair it with an entry rule from a different category, such as a breakout or a pullback trigger.

Frequently asked questions

What ADX level indicates a strong trend?

Conventionally above 25, with readings above 40 indicating a very strong trend. These thresholds vary by instrument and timeframe, so the more robust approach is to compute the historical distribution of ADX for your market and use percentiles, for example the top third of readings, rather than fixed numbers.

Does ADX tell me whether to buy or sell?

No. ADX measures the strength of a trend without regard to direction, so it rises in both strong rallies and sharp declines. Direction must come from the +DI and -DI comparison, from a moving average, or from price structure. Using ADX alone as a directional signal is a fundamental misreading.

How do I use ADX with other indicators?

As a gate rather than a signal. Enable trend-following entries only when ADX confirms a trend exists, and enable mean-reversion entries only when it confirms one does not. This pairing addresses the main weakness of each family and is more valuable than adding a second entry indicator.

Is ADX a lagging indicator?

Yes, and noticeably so, because it is a smoothed average of an already smoothed calculation. It confirms that a trend is in place rather than anticipating one. That lag is acceptable for its main use as a regime classifier, and unacceptable if you try to use it for entry timing.

What is the best ADX setting?

The default 14 periods is appropriate for most uses. Shorter periods make the indicator responsive enough to lose its filtering value; longer periods delay regime recognition further. As with other indicators, prefer a setting that works across a range rather than the one that performed best historically.

Test this idea before you trade it

Describe the rules in plain language and AlgoTrader AI turns them into a structured strategy blueprint with a configurable historical backtest, cost assumptions, and exportable code.

Build a backtest

Keep reading

Referenced by

Educational use only. This guide explains how a strategy works. It is not investment advice, not a recommendation, and no result described here is a forecast. Test any approach on historical and out-of-sample data, size positions conservatively, and never risk money you cannot afford to lose.