The ADX measures how strong a trend is, not which way it is going, so day traders use it as a gate rather than a signal: take the setup when the ADX says there is a trend to trade, stand aside when it says there is not. Direction comes from something else.

The settings are where it gets messy. Published ADX periods for day trading run from 2 to 30, and the level that counts as a trend moves too. A borrowed number can behave differently for three reasons: the ADX is two inputs rather than one, its smoothing has no fixed lookback, and the threshold moves with the chart timeframe.

What does the ADX measure, and does it show direction?

Trend strength on a scale of 0 to 100, and no, it does not show direction. Capital.com states the limit in one line: "It doesn't indicate the direction of the trend, but rather its vigour." Elsewhere it says what you need instead: "You need +DI and -DI (or other indicators) to determine if it's an uptrend or downtrend." AquaFunded puts it as measuring "the strength of directional price movement, not the direction itself".

The claim does get stated both ways in the wild. The Secret Mindset opens by calling the ADX "an indicator that measures trend strength and shows trend direction", then says further down that "In other words, the ADX is non-directional". The second version is the right one.

The construction runs in two stages, and Capital.com publishes both. The raw inputs are smoothed first, "over a period, usually 14 periods". Then the index built from them is smoothed again, "a smoothed average of the DX over the chosen period".

Is the ADX one setting or two?

Two, and most settings advice sets only one of them. Stock Pathshala names both inputs: "Just put 3 in ADX smoothing and DI length."

That follows from the two stages: one number smooths the directional movement inputs, the other smooths the index built from them. So if your platform shows two boxes and the setting you found gives one number, you have not been told which box it goes in. Write down both, or a later test will not be comparable.

What ADX period do day traders use?

Anything from 2 to 30, depending on whose page you landed on.

Period	Where it is published	Offered as
2	Capital.com	A named strategy to "identify extremely strong, short-term trends or exhaustion points"
3	Trading Strategy Guides, Stock Pathshala	"The best ADX setting for day trading is the 3-period", because "it increases the sensitivity of price movement"
5 or 8	Capital.com	"Use shorter ADX settings, like 5 or 8 periods, to increase sensitivity"
7	Stock Pathshala, Trading Strategy Guides	For a 5-minute chart: "Put 7 in the ADX smoothing and DI length"
14	Capital.com, Stock Pathshala, StockGro, AquaFunded	Wilder's default, described as "usually 14 periods" and "a default 14-bar smoothing window"
20	The Secret Mindset	Used throughout its worked construction: "a moving average of the DX over 20- period"
3 to 30	Stock Pathshala	A range, not a number: "the ADX setting for day trading varies from 3 to 30"

Capital.com alone carries four of them, for four different purposes, and StockGro says to shorten the default without naming a replacement, since "The day traders usually shorten the default 14-period setting for quicker responses." There is nothing to average here.

Does a shorter ADX period mean a shorter lookback?

No, and the common way of putting it is wrong. Stock Pathshala converts periods into clock time: "If we have a 5-minute chart, then ADX for 14 periods shows the data of 70 minutes [ 14*5 = 70 ]" and "ADX of 3 periods will show data of the past 15 minutes."

Capital.com publishes the formula that settles it, which written on one line is:

Current ADX = ([previous ADX X 13] + current DX) / 14

Every ADX value is built from the previous one, which was built from the one before, back to the start of the series. No bar ever drops out. A short period makes recent bars count for much more and old ones count for very little, and very little is not nothing, so the reading is not a window over the last so many minutes.

There is no clean clock equivalent to put in its place. What it does change is the warm-up: a recursive average needs a long run of bars before it settles, so the first stretch of any test window is not comparable with the rest. Start the sample well after the first bar the indicator plots.

What does an ADX above 25 mean?

A trend worth taking a setup in. 25 is the level you will meet most often, and what sits around it is not agreed at all.

Capital.com publishes a graded scale: "A move above 20 often signals the start of a trend"; "ADX below 20-25 indicates a weak or ranging market"; "ADX rising above 25 can represent an emerging or strengthening trend"; "ADX above 40 signifies a strong, well-established trend". Its 2-period strategy wants far more, above "60 or 70", and its moving-average-cross strategy uses "ADX is above 30 and rising".

Fisdom publishes four bands with different boundaries: "The ADX indicator between 0-20 indicates a non-trending stock, the ADX indicator between 20-45 indicates a strong trend, the ADX indicator between 45-60 indicates a very strong trend while an indicator over 60 indicates an extremely strong trend."

On Fisdom's scale a strong trend starts at 20. On Capital.com's, 20 is roughly where a weak or ranging market ends and a well-established trend needs 40. A rule that says "trade when the trend is strong" does not survive the move from one scale to the other.

One rule set can even disagree with the next on the same page. Trading Strategy Guides uses 25 at both ends of one strategy, above 25 to enter and "breaks back below 25" to take profit, while another looks for clues "when ADX falls below 25" and a third waits "until the ADX breaks above 50".

Does the level change with the chart timeframe?

One published ladder says yes, moving the period and the level together. Trading Strategy Guides answers it in its own FAQ:

1-minute chart: "some commonly used settings are 3 for the period, and 10 for the threshold level."
5-minute chart: "some commonly used settings are 7 for the period, and 20 for the threshold level."
15-minute chart: "some commonly used settings are 14 for the period, and 25 for the threshold level."

Each answer opens by saying it depends "on the trader's preferences and trading style", so these are offered as common rather than correct. The ladder also puts the same page's headline figure of 3 in context: that is the 1-minute answer rather than a general one. The ladder varies by timeframe only. Whether 25 means the same on gold as on EUR/USD, or on an index as on a currency pair, is not a question any published setting answers, so a level for the instrument you trade is one you have to measure.

Is the ADX leading or lagging?

Lagging, wherever it is addressed. Capital.com is explicit about the cost: "the ADX is a lagging indicator. It is based on historical price data and confirms a trend only after it has already begun. This means traders using ADX might miss the very early stages of a new trend."

Fisdom adds what that means in a fast market: it "is generally slow to respond to dynamic market changes and therefore may not reflect the accurate market conditions." AquaFunded ties it back to the settings: "smoothing results in a lagging read. Because of this, ADX works best with rules that allow a slight delay for fewer false signals." The Secret Mindset lists it as a "lagging indicator, that follows the price".

A shorter period reacts sooner and flips more often, a longer one confirms later and flips less, and neither removes the lag, because the smoothing is what creates it.

How do you read +DI and -DI against the ADX level?

The DI lines carry the direction, the ADX carries the strength, so the usual reading pairs a crossover with a level.

Capital.com defines them: +DI "measures the strength of upward price movement", -DI "measures the strength of downward price movement". Capital.com also describes the shapes: "In a strong uptrend, +DI will be significantly above -DI, while in a strong downtrend, -DI will be significantly above +DI." In a range, "+DI and -DI will often be close to each other, reflecting the indecision in the market."

StockGro describes the pairing as a rule: "The traders might enter a long position when +DI stays above -DI and ADX climbs over 25", with the mirror for shorts when "-DI leads and ADX rises above 25", and an exit "if the opposite crossover occurs".

Capital.com applies the same idea to breakouts: "A breakout that occurs with a rising ADX above 25 is generally more reliable than one that occurs with a low or falling ADX, as it indicates genuine underlying strength in the new trend." Those are the published readings, not recommendations from us, and we make no claim about whether any of them works.

How do you turn an ADX filter into a rule you can test?

Write it so it has exactly one reading, which takes four decisions rather than one: which period, in which of the two boxes, at which level, and whether that level holds on every timeframe and every symbol you trade. "Only trade when the ADX confirms a trend" is not yet a rule.

A filter is also the easiest kind of rule to fool yourself about, because it is defined by the trades it removes, and those are invisible unless you run the same sample twice, once with the filter on and once with it off.

In ChartLabs you replay historical data candle by candle and place each trade by hand, so a filter gets applied in the moment rather than in hindsight. The data is sourced from Dukascopy, tick-level where it is available, and cleaned to remove anomalies. The ADX is in the Indicators menu, inside Directional Movement Index, which draws +DI, -DI, ADX and ADXR together in their own pane below price, so the crossover and the level are one reading rather than two separate add-ons. If you want a version whose two smoothing inputs you set yourself, write one in the built-in Pine Script editor, which uses Pine Script v5/v6 syntax. The Pine Scripts tab has two Examples, "Example: Line on price" and "Example: Mark bars by a rule", and clicking one opens a copy in the editor to start from. Pine there draws, it does not place trades, so the testing stays yours, bar by bar. Sub-panel indicators save themselves as you work, so the setup you tested is the setup you come back to.

Keep one untouched stretch of history for last and run the filter comparison on that. Tuning a level on the same data you then judge it with gives you a number that means nothing.

Where to go next

For the process rather than the indicator, backtesting a trading strategy covers writing rules down and testing them. Forex backtesting covers why session and pair behaviour change results, which is the same reason one ADX level is unlikely to travel between symbols. And manual versus automated backtesting covers what a by-hand test tells you that a report does not.

One plan, $20 a month or $200 a year, with a 7-day free trial and no credit card required: start the free trial.

How to Use the ADX Indicator for Day Trading