ADX Rising in a Downtrend: Trend Strength vs Direction

A falling market and a rising ADX line are not arguing. ADX measures trend strength without giving that trend a direction. A persistent decline can therefore produce a rising reading just as a persistent advance can.

That distinction matters in TradingView automation. If a rule turns “ADX increased” into a buy instruction, it has quietly substituted strength for direction. Before an alert becomes an execution command, define the directional condition separately from the strength filter. A connector cannot repair a trading rule that asked for the wrong side.

Read the three lines as two different questions

TradingView's ADX documentation distinguishes ADX from the accompanying directional indicators. Its standalone ADX indicator does not show both DI lines; add Directional Movement Index (DMI) to inspect all three together.

• ADX: how strong is the directional tendency measured by this indicator?
• +DI above -DI: positive directional movement dominates the comparison.
• -DI above +DI: negative directional movement dominates the comparison.

These are smoothed measurements, not a promise about the next candle. In particular, falling ADX does not itself mean falling price. An uptrend can lose strength while price continues higher.

The absolute value removes the sign

The useful piece of the calculation is DX = 100 × |+DI − -DI| / (+DI + -DI), before smoothing produces ADX. This expression assumes the denominator is positive. The absolute value is why swapping the two DI readings leaves the strength calculation unchanged.

Consider invented readings for illustration: +DI is 10 and -DI is 30. DX is 100 × 20 / 40 = 50. Swap them, so +DI is 30 and -DI is 10, and DX is still 50. The directional balance reversed; its normalized separation did not. This is a calculation example, not an observed trade or an ADX forecast.

Because ADX smooths DX, a single pair of DI readings is insufficient to reconstruct the current ADX. Comparing today's price change with today's ADX change also compares two different things: one bar's price movement and a measure carrying information from earlier bars.

A threshold is a filter, not a trade direction

TradingView's DMI guide describes the familiar 20 and 25 reference levels, while warning that useful thresholds depend on the instrument. Its DMI settings separate DI length from ADX smoothing, with defaults of 14 for each. Record both when comparing charts or tests.

“ADX above 25” and “ADX rising” are different filters. A move from 17 to 19 meets the rising test but remains below 25. A move from 33 to 31 fails the rising test but remains above 25. Neither condition selects long or short on its own.

Before adding another filter to a Supertrend rule, write down what it changes. Requiring high ADX could exclude entries during weaker measured trends; requiring rising ADX could exclude entries when strength is easing. Whether either improves results is an empirical question, not part of the indicator's definition.

Turn the idea into an unambiguous alert rule

Here is one hypothetical specification for testing, not a recommended strategy: evaluate completed bars; permit a long candidate when +DI is above -DI and ADX is above 25; permit a short candidate when -DI is above +DI and ADX is above 25. Treat equal DI readings as neither direction. Define exit and sizing rules separately.

Now choose the event. “The long candidate is true” describes a state that could persist for ten bars. “The long candidate became true on this bar” describes a transition. If you want one entry when the combined condition first becomes true, compare it with the previous bar's combined condition and specify how an existing position blocks or permits another entry.

Suppose +DI crossed above -DI yesterday while ADX was 23, then ADX reaches 26 today with +DI still higher. A rule requiring a fresh DI crossover and ADX above 25 on the same bar would not enter today. A rule using the transition into the combined eligible state could. Neither interpretation is implied by the phrase “use ADX confirmation.” It must be written down.

Use the strategy alert timing guide to keep the signal event distinct from simulated fills and execution. During a test, record bar time, both DI values, ADX, the previous eligible state, the chosen direction and position state. This makes it possible to explain why an alert did or did not fire without inventing a story from the chart afterward.

The practical check is simple: if the strategy can explain its strength filter but cannot explain its direction, it is not ready to send orders. A rising ADX in a falling market is often perfectly consistent. The buy instruction is the part that needs questioning.