ADX Indicator — Measuring Trend Strength and Filtering Weak Signals

There's a whole family of tools in trading that lose as standalone strategies and win as filters. ADX is their best example — with the numbers on the table. Tested solo on a US stock index it produced — per QuantifiedStrategies — a CAGR of about 6.5% (1993–2022). Nothing to write home about. But the same indicator, bolted on as a filter to a contrarian strategy, did something more interesting than lifting the return: it doubled the average profit per trade (from 0.54% to 1.1%) and cut time in the market from 24% to 15%.

Read that again: the strategy was in the market less than two-fifths of the time and still beat the full-exposure result. ADX didn't earn that money — ADX just refused to let it be spent in bad conditions. This whole article is about that difference.

What Is ADX

ADX (Average Directional Index) was developed by J. Welles Wilder — the same engineer who gave traders RSI and ATR — and published in 1978. The indicator consists of three lines:

• DI+ (Directional Indicator plus) — measures upward pressure: how often and how strongly the market makes new highs,
• DI− (Directional Indicator minus) — measures downward pressure the same way,
• ADX — a smoothed (standardly 14-period) measure of the difference between DI+ and DI−, rescaled to a 0–100 range.

The intuition without formulas: if the market consistently moves one way — either way — one DI line dominates the other and ADX rises. If the market whips around in place, DI+ and DI− keep crossing and ADX falls. Hence the most important property, on which all correct use depends: ADX measures the strength of a directional move, not its direction. A rising ADX in a bear market means a strong bear market, not a bounce.

Conventional interpretation thresholds:

• ADX < 20 — no trend, a sideways or choppy market,
• 20–25 — a transition zone,
• ADX > 25 — a trend is present and tradable,
• ADX > 40 — a very strong trend (and often already mature).

One common mix-up is worth defusing right away: ADX doesn't measure volatility — it measures directionality. A market can have huge daily swings (high ATR) and a low ADX at the same time, because it's whipping around a wide range with no net progress. And conversely, a calm, steady march in one direction produces a high ADX at moderate ATR. These are two different questions — "how hard is the market moving" and "is it actually going somewhere" — and it's the second one that matters for filtering trend strategies.

What the Numbers Show

Two measurements from QuantifiedStrategies, placed side by side, tell the whole story (sources: ADX alone and paired with RSI, ADX as a filter; as of 09.2026):

ADX as a standalone strategy: CAGR ~6.5% on 1993–2022 data. For comparison, simply buying and holding the index returned about 10.5% a year over a similar period. An indicator meant to measure a trend didn't, on its own, generate an edge over just owning the market.

ADX as a filter on a contrarian strategy (S&P 500): the annual return fell from 9% to 7%, but time in the market dropped from 24% to 15%, the average profit per trade rose from 0.54% to 1.1%, and the profit factor from 2.15 to 2.4. In a separate test, a pair of indicators from the RSI/ADX/IBS trio produced 270 trades on SPY with an average profit of 0.86% at just 14% time in the market. The mechanics of this duo: RSI flags moments of short-term oversold conditions (opportunities to enter in the spirit of mean reversion), and ADX classifies the conditions in which such a contrarian entry makes sense — cutting out the market phases where catching a falling knife ends in getting run over.

Why is time in the market the most important number here? Because the return is calculated on capital that carried no market risk 85% of the time. A lower annual return with far less time in the market can be better on a risk-adjusted basis than the raw CAGR figures suggest — and the idle capital can be working elsewhere in the meantime. This is exactly the property worth paying for in trading: being in the market rarely, but under the right conditions.

A structural conclusion, not just a numerical one: ADX belongs to the category of regime classifiers, alongside price's position relative to a long-term average or the width of an EMA ribbon. A regime classifier doesn't say "buy now" — it says "your strategy historically worked in these conditions, and didn't in those." That's less exciting than signals. And much more valuable.

How to Trade It — Step by Step

Two honest variants — in both, ADX is a filter, never a trigger.

Variant A — a filter for a trend strategy:

1. Take your signals from your own strategy — e.g., a breakout, a pullback to an EMA ribbon, a moving-average crossover.
2. You only get the green light at ADX > 25 (calibrate the threshold on your own data). A rising ADX means the trend is gaining strength, and continuation signals have the wind at their back.
3. ADX < 20 means the trend strategy sleeps. Don't go looking for "smaller trends on a lower timeframe" — that's the most common loophole for overtrading a sideways market.

Before picking a variant, write down the overarching rule: ADX never opens a position — ADX only has the right to veto one. This asymmetry protects against the most common abuse of the indicator: treating a rising line as an invitation to buy.

Variant B — a filter for a contrarian strategy (the equivalent of the tested RSI+ADX):

1. Entry signal: short-term oversold conditions, e.g., from RSI or the stochastic.
2. The ADX filter decides whether the bounce is tradable. Work out the threshold logic empirically: in stock markets contrarian entries often worked precisely in combination with an ADX filter — but whether a low-ADX or a high-ADX requirement is better depends on the market and needs its own test. Don't guess — measure.
3. A stop and a time-stop are mandatory — a contrarian trade without a defined exit isn't a strategy, it's an opinion.

A numeric example on BTC (illustrative, Variant A): the daily timeframe, a $10,000 account, 0.75% risk = $75. Your base strategy gives a long signal on a pullback to the EMA 21 at a price of $64,000. You check the filter: the daily ADX(14) reads 31 and is rising, with DI+ clearly above DI− — the entry is cleared. Stop below the local low of the correction at $61,800 ($2,200 of risk per BTC). Position size = 75 / 2,200 ≈ 0.034 BTC. A month later an identical price signal appears with ADX at 16 — and here's the whole job of the filter: there is no trade. Half of ADX's value lies in the trades you never take.

When It Doesn't Work and Common Mistakes

• ADX as a signal generator. The standalone results (a 6.5% CAGR) speak for themselves. The indicator describes the market's character with a lag — entries from ADX alone are late to the start of a trend and overstay its end.
• Confusing strength with direction. A rising ADX during a decline confirms the bear market. Buying "because ADX is rising" without checking DI or price structure is a page-one manual mistake, and a surprisingly common one.
• DI+/DI− crossovers as a system. Wilder's classic signal produces a string of false crossings in sideways markets — exactly where ADX is low, which is exactly where the indicator itself is warning you not to trade. Using DI crossovers against a low ADX ignores the better half of the tool.
• The 20–25 zone isn't a signal, it's fog. A market moving in and out of the transition zone can flip the filter on and off every few candles. Fixes: hysteresis (turn on at 25, turn off at 20) or requiring the reading's direction to be confirmed for 2–3 candles.
• A peak in ADX ≠ a peak in the trend. A falling ADX after a high reading means the momentum is fading, not reversing — the trend often continues at a calmer pace. Closing good positions "because ADX is falling" means giving up the tail of profits that trend-following strategies live on.
• One threshold for every market. The textbook "25" comes from stock markets decades ago. BTC on the 4H, gold on the daily, and a tech stock on the H1 all have different ADX distributions — a threshold without calibration on your own data is a number pulled out of thin air.

No-hype summary: ADX is a weak advisor and a good bouncer. It doesn't tell you what to buy or when — it tells you whether the conditions in the room favor your strategy, and it consistently throws you out of parties where you've historically lost money. The numbers show that exactly this boring role — cutting exposure down to periods with an edge — doubled the base strategy's average profit per trade. Don't look to ADX for signals. Look to it for the courage not to trade.

FAQ

Rafał — Krypto Bez Ściemy

Trader and founder of Strefa Tradingu. He’s been breaking crypto down on YouTube for years — no hype, no signals, with a focus on market structure and risk management.

kryptobezsciemy.plInstagramYouTubeKlub Zero