OptiNod Academy

ADX / DMI - Measuring Trend Strength

Use it as an entry filter: below ADX 20, discard all other trend-indicator signals.

Many articles present ADX as an entry tool: buy above ADX 25, sell below 25, or enter when +DI crosses above -DI. That is not how Welles Wilder intended ADX to be used when he introduced it in *New Concepts in Technical Trading Systems* in 1978. Wilder designed ADX as a filter for identifying the current market regime, and wrote that trade direction should be determined with other tools.

That distinction matters. ADX loses meaning if you read it as a simple standalone value; it should be treated as a condition. ADX below 20 means "there is no trend in this market right now." It is not a sell signal. If you take trend-indicator signals such as a MACD golden cross or a 200 EMA breakout when there is no trend, they will keep producing false signals. Used as an entry filter, ADX removes more than half of trend-indicator signals before they ever reach the execution stage. That is where it removes the largest source of losses from a trend-following system.

**ADX is the gate for trend signals**Below 20, discard trend-indicator signals. Above 20, evaluate other entry triggers.

Asymmetric Response - Sharp at the Start, Slow at the End

ADX is built by first smoothing +/-DM over 14 bars, normalizing them by TR (true range) to get +DI and -DI, then calculating DX = 100 x |+DI - -DI| / (+DI + -DI), which divides the distance between the two lines by their sum. ADX is that DX value smoothed again over 14 bars using Wilder's smoothing. In other words, it is a smoothed ratio of the absolute difference between +DI and -DI divided by their sum. Because of that structure, ADX responds asymmetrically to trend starts and trend endings.

At the start of a trend, +DM expands quickly on one side while the other side stays near zero. The ratio rises sharply. ADX may move between 15 and 18 in a range, then break above 25 within 3 to 5 bars once a real trend begins. That speed is what makes ADX valuable as an entry filter.

At the end, the opposite happens. Even as the trend starts to cool, +DM and -DM both fade gradually, and the ratio flattens out. ADX usually takes 10 to 20 bars to fall from 40 to 25. Because of this asymmetry, ADX is late at signaling trend endings. By the time the ending is obvious, price has already retraced substantially.

That is why ADX should have different roles in entries and exits. At the entry stage, ADX is fast and useful: the sharp break through 20. At the exit stage, ADX is slow, so exits should come from other tools such as a break of the prior swing in price structure or an ATR trailing stop.

**Fast at the start, slow at the end**ADX rises quickly when a trend begins, but its exit signal is late, so price structure should handle exits.

Entry Filter - Use It as a Gate for Other Signals

The cleanest way to use ADX as an entry gate is to place one simple condition in front of every trend-indicator signal. If ADX is not above 20, discard every trend signal on that bar.

> On the MSFT daily chart, price is above the 200 EMA and pulling back toward prior support.

> Consider a long entry at the close of the bar where RSI reclaims 50.

> But if ADX on that same bar is below 20, skip the entry entirely.

> Enter long only when ADX is above 20, and place the stop below the pullback low.

> If ADX falls below 20 while the position is open, treat the trend as gone and take partial profits.

This single gate removes nearly half of the false signals from a trend-following system. Even a large asset such as SPY spends roughly 40% of the year below ADX 20. Any "trend signal" that appears during that zone is likely noise from the start.

The common trend threshold is 25. This article uses 20 as the gate because it lets early trend candidates through before ADX reaches 25. When a setup requires higher confidence, use 25 or 30 as an additional reference.

Conversely, when ADX is above 30, the reliability of the same indicator signal clearly improves. The signal itself has not changed; the trend condition has.

**Reject weak signals at the first gate**Checking for ADX above 20 and rising filters out false range-bound trend signals early.

Falling ADX Above 25 - Weakness Before Price Shows It

There is one place where ADX direction carries more information than the value itself: when ADX peaks in the 25 to 40 zone and starts to fall.

Price may still be moving in the trend direction and may even make another new high. But a falling ADX means the spread between +DI and -DI, or DX, is narrowing. That is a direct signal that the incremental buying pressure needed to make new highs is weakening. The force behind new highs is fading.

In the strong rally in copper futures in April 2024, daily ADX rose to 42 and then started to fall. Price made two more attempts at new highs afterward, but ADX never recovered. After the second failed new-high attempt, a large six-week correction followed. The ADX peak came 9 bars before the price peak.

This is a signal to take profits sooner on an existing position. Treat it as profit management on an open trade, and keep new entries out. Once ADX starts falling above 25, stop taking new trend entries and close half of the open trend position. Trail the remainder using price structure until the prior swing low breaks.

**Strength can roll over before price does**Even if price makes one more high, falling ADX means stop new entries and shift to profit management.

+DI/-DI Crosses - Do Not Use Them Alone

Some beginner material teaches the crossover of the two lines plotted with ADX, +DI in green and -DI in red, as a buy or sell signal. That use differs from Wilder's original text, and it also breaks down cleanly in backtests.

Across all market states, including ranges, a simple +DI/-DI crossover has a reliability near 50%. It is close to a coin toss. In ranges, the two lines cross frequently and produce a stream of false signals. A crossover only has meaning during a trend phase: ADX at 20 or higher and rising. If either condition is missing, discard the crossover.

With that gate in place, the +DI/-DI crossover becomes a confirmation signal that the trend is accelerating. Do not treat it as a standalone entry trigger. Use it as secondary confirmation alongside another signal. Using it alone is the opposite of ADX's original purpose.

**+DI/-DI crosses need a trend phase**Low-ADX crosses are close to a coin toss. Use them for directional confirmation only when ADX is rising.

ADX Above 50 - It Does Not Stay There Long

ADX above 50 is statistically short-lived. On the SPY daily chart, ADX above 50 lasting more than 5 bars at a time happens only once or twice a year. Even in high-volatility assets such as altcoins, 5 to 10 bars is usually the limit.

The most damaging action in this zone is adding to a position. If you read ADX 50 as "the trend is very strong" and enter, you are likely entering the final leg of the trend at the worst reward-to-risk. Above 50, the correct response is to hold off on new entries, take profits on a larger portion of existing positions, and raise the trailing stop once the ADX peak is confirmed.

On the daily chart of silver (SLV), ADX reached 53 during the strong rally in May 2024, then peaked and started falling after 4 bars. Price rose another 3% after the peak, but then corrected 18% over the next month. ADX above 50 looks like an opportunity zone, but in practice it is a danger zone.

**Treat ADX above 50 as exhaustion**The longer ADX stays above 50, the more you should avoid new entries and respond with partial exits and higher trailing stops.

Why 14 Is the Smoothing Balance Point

Many traders reduce Wilder's default 14 period to 7 or 10 to get faster signals. That change breaks the calculation balance of ADX.

The ADX formula itself contains two layers of 14-bar smoothing. The first smooths +DM and -DM. The second smooths the ratio. When both smoothing layers are aligned at 14 bars, noise is filtered well enough while real trend starts still register within 3 to 5 bars. If you reduce the period to 7, noise passes through, and ADX jumps between 15 and 25 almost every bar. At that point, it loses value as an entry filter.

If you need a faster signal, do not shorten the ADX period. Move down one timeframe instead. Daily ADX(14) and 1-hour ADX(14) preserve the same smoothing balance while looking at a shorter time unit.

Splitting Timeframes to Offset Lag

ADX is a smoothed value of a smoothed ratio, so it is structurally a lagging indicator. By the time ADX closes above 20, the trend has already progressed to some degree. To reduce that lag, split the analysis across timeframes. Shortening the period is not the answer because it breaks the smoothing balance.

Use daily ADX to decide whether a trend exists now. It answers: "Is there a trend, and is that trend accelerating or cooling?" Entry timing should come from the price structure of the same asset on the 1-hour or 4-hour chart. While daily ADX confirms a trend phase at 22, enter on the 4-hour bar where price pulls back to prior support and recovers.

Handled this way, ADX's lag becomes a deliberate confirmation mechanism for trend state. The shorter timeframe supplies the faster entry signal, while daily ADX slowly confirms that the entry sits inside a real trend phase. Once the two timeframes have separate roles, lag changes from a weakness into a strength.

Daily ADX confirms the trend while a shorter timeframe times entry on a pullback to support

ADX Needs Other Tools Around It

An ADX-based system does not work properly unless you include two other components.

Direction tool: ADX only tells you strength. Direction comes from tools such as whether price is above or below the 200 EMA, or the sequence of highs and lows in price structure. ADX above 25 only says "there is a trend." Another tool must answer whether it is up or down.
Exits from price structure: ADX is slow at signaling trend endings. Trend exits should therefore come from price-structure signals such as a close below the prior swing low. Use ADX only to bring profit-taking forward, and leave the exit trigger to price structure.