What Is the Directional Movement Index (DMI)? Formula and Trading Signals

Search Fintrev

Search our guides, reviews and glossary.

The directional movement index (DMI) is a technical indicator that measures whether buyers or sellers have been pushing price further over a set number of periods, and by how much relative to the market’s normal range. It plots two lines: the positive directional indicator (+DI), which tracks upward movement, and the negative directional indicator (-DI), which tracks downward movement. J. Welles Wilder Jr. introduced it in 1978 in New Concepts in Technical Trading Systems, together with the average directional index (ADX). The ADX is derived from the two DI lines and measures trend strength without regard to direction.

In practice, “DMI” refers to one of two things. Sometimes it means only the +DI and -DI pair. Sometimes it means the full three-line system that includes the ADX. Many charting platforms show all three lines under one DMI indicator, so check which version a source means before you apply its rules.

What the DMI measures

The core idea is directional movement: the part of today’s range that extends beyond yesterday’s range. If today’s high is above yesterday’s high by more than today’s low is below yesterday’s low, the bar counts as upward directional movement. If the low extends further, it counts as downward movement. Bars that stay inside the prior bar’s range contribute nothing in either direction.

Each day’s directional movement is then smoothed and divided by the smoothed true range. The result tells you how much of the market’s recent movement has been directional. A +DI of 30 means smoothed upward directional movement equals 30% of smoothed true range over the lookback period. It does not mean price rose 30%.

The DMI formula

Wilder’s default lookback is 14 periods, and the steps below use it. The same arithmetic works on any timeframe, from one-minute forex charts to weekly stock charts.

Directional movement (+DM and -DM)

For each period, calculate two raw differences:

• Up move = current high − previous high
• Down move = previous low − current low

Then apply Wilder’s rules:

• +DM = up move, if the up move is greater than the down move and greater than zero. Otherwise +DM = 0.
• -DM = down move, if the down move is greater than the up move and greater than zero. Otherwise -DM = 0.

At most one of the two can be positive on any bar. An inside bar (lower high and higher low) produces zero for both. An outside bar (higher high and lower low) produces a value only on the side with the larger extension.

True range

The true range (TR) is the largest of:

1. Current high − current low
2. |Current high − previous close|
3. |Current low − previous close|

Including the previous close captures gaps, which a simple high-minus-low range would miss. The same TR feeds the average true range (ATR), and Wilder used it here as the denominator so the DI lines are normalized for volatility.

Wilder smoothing

Wilder did not use a simple average. For the first value, sum the first 14 readings of TR, +DM and -DM. Each subsequent value is:

Smoothed value = prior smoothed value − (prior smoothed value ÷ 14) + current value

This is a running sum that gradually forgets older data. It behaves much like an exponential moving average with a smoothing factor of 1/14. Because +DM, -DM and TR are all smoothed the same way, it does not matter whether you work with smoothed sums or smoothed averages. The ratio between them is identical.

+DI, -DI, DX and ADX

• +DI = (smoothed +DM ÷ smoothed TR) × 100
• -DI = (smoothed -DM ÷ smoothed TR) × 100
• DX = (|+DI − -DI| ÷ (+DI + -DI)) × 100
• First ADX = average of the first 14 DX values
• Later ADX = ((prior ADX × 13) + current DX) ÷ 14

The DX measures how lopsided the two DI lines are. When one dominates, DX is high. When they are close, DX is low. The ADX is a smoothed DX, so it moves more slowly and is easier to read. Because of the two layers of smoothing, a 14-period ADX needs roughly 27 or 28 bars of data before it produces its first value, and early readings settle as more history accumulates.

Worked example (hypothetical)

The figures below are invented to show the mechanics on a EUR/USD daily chart. They are not market data.

Day 1: high 1.1050, low 1.0990, close 1.1030 Day 2: high 1.1080, low 1.1010, close 1.1070

Step 1, directional movement for Day 2:

• Up move = 1.1080 − 1.1050 = 0.0030 (30 pips)
• Down move = 1.0990 − 1.1010 = −0.0020, which is negative
• So +DM = 0.0030 and -DM = 0

Step 2, true range for Day 2:

• High − low = 0.0070
• |High − previous close| = |1.1080 − 1.1030| = 0.0050
• |Low − previous close| = |1.1010 − 1.1030| = 0.0020
• TR = 0.0070 (70 pips)

Step 3: assume that after 14 days of this process, the smoothed values are smoothed +DM = 0.0180, smoothed -DM = 0.0090 and smoothed TR = 0.0900.

• +DI = (0.0180 ÷ 0.0900) × 100 = 20
• -DI = (0.0090 ÷ 0.0900) × 100 = 10
• DX = (|20 − 10| ÷ (20 + 10)) × 100 = 33.3

In this example, upward directional movement is twice the downward movement. The DX of 33.3 would then be averaged with other DX readings to build the ADX. A single strong day changes the smoothed values only modestly, which is why the DI lines react gradually rather than jumping with every bar.

How to use the DMI in trading

Crossovers

The basic DMI signal is a crossover:

• Bullish: +DI crosses above -DI. Upward directional movement now outweighs downward movement, and an uptrend may be starting. A trader using this rule would close shorts and consider a long.
• Bearish: +DI crosses below -DI. Downward movement now dominates, and a downtrend may be starting. Longs would be closed and a short considered.

A crossover reflects what has already happened over the lookback period. Price can cross, move a little and then reverse, so the signal on its own is a prompt to look closer rather than a complete entry rule.

Trend confirmation

Many traders use the DMI to check signals from other methods rather than as a trigger. When +DI sits well above -DI, the case for a long position from a breakout or pattern is stronger. When -DI is well above +DI, the evidence favors short positions or staying out of longs. A wide gap between the lines means one side has clearly dominated recent movement. Lines that sit close together and keep touching describe a market without a directional edge.

Adding the ADX filter

The ADX answers a separate question: is there a trend worth trading at all? A rising ADX means directional movement is becoming more one-sided, in whichever direction the DI lines show. A falling ADX means the trend is losing force or the market is moving into a range. Traders often treat ADX readings above 20 or 25 as trending and readings below that as non-trending. These are conventions, not fixed laws, and the right threshold depends on the market and timeframe.

A common combined approach is to take DI crossovers only when the ADX is above the chosen threshold and rising, and to ignore crossovers when the ADX is low. This filters out many of the whipsaw signals that occur in choppy conditions. The trade-off is that some early entries at the start of new trends get skipped, because the ADX often lags the move.

Note that the ADX rises in strong downtrends too. A high ADX with -DI on top describes a strong bearish trend, not a bullish one.

Wilder’s extreme point rule

Wilder added a refinement to reduce false crossovers. On the day the DI lines cross, note that bar’s extreme price: the high for a bullish cross, the low for a bearish cross. Price must then trade beyond that level to confirm the signal. If it does not, and the lines cross back, the signal is treated as a failure. The crossover bar’s opposite extreme can also serve as a reference point for a stop-loss.

Choosing settings

Fourteen periods is the standard, but the setting is adjustable:

• Shorter periods (for example 7 or 10) make the DI lines react faster and cross more often. That means earlier signals but more false ones.
• Longer periods (for example 20 or 30) smooth the lines and reduce crossovers. That means fewer signals that arrive later.

Many platforms let you set the DI length and the ADX smoothing length separately. Implementations can also differ in their smoothing method, so readings may not match exactly between two charting packages. If you are building a scan or strategy, use the formula the platform documents rather than assuming it matches Wilder’s arithmetic.

Limitations of the DMI

• It lags. Every input is a past high, low or close, and the results are smoothed twice for the ADX. Turning points show up after the fact.
• Whipsaws in ranges. In sideways markets the DI lines can crisscross repeatedly, producing a series of signals with no follow-through. This is the main reason traders pair it with the ADX or a higher-timeframe trend filter.
• No price targets. The DMI says which side has dominated recent movement. It gives no estimate of how far price will go or where to exit.
• Sensitivity to gaps and spikes. A single large bar can dominate the smoothed values for several periods, especially on thin markets or around news releases.
• No guarantee of follow-through. A crossover can occur and price can still fail to trend, turning the trade into a loss.

These limits are one reason to size positions as if any given signal could fail. A position size calculator can turn a stop distance taken from the chart into a lot size that caps the loss per trade.

DMI vs. similar indicators

DMI vs. ADX

The two are parts of one system, not alternatives. The DMI’s +DI and -DI give direction. The ADX gives strength and is calculated from the DI lines. Comparing “ADX vs. DMI” really means comparing a strength reading with a direction reading. Most traders who use one look at both.

DMI vs. Aroon

The Aroon indicator, developed by Tushar Chande in 1995, also aims to identify trends and their strength, but it is built differently:

• DMI compares consecutive highs and lows and smooths the size of the extensions. It is usually read through DI crossovers and the ADX, and it is suited to confirming direction and strength in markets that are already trending.
• Aroon measures how many periods have passed since the highest high and the lowest low in the lookback window. It is usually read for the start and end of trends and for spotting consolidation.

The DMI cares about how much price extended. The Aroon cares about how recently it made a new extreme.

DMI vs. MACD and Parabolic SAR

MACD gauges trend direction and momentum from the gap between two exponential moving averages of closing prices. The DMI uses highs, lows and true range instead. The Parabolic SAR, another Wilder indicator, plots trailing stop levels that flip when price reverses. Traders sometimes pair it with the DMI, using the DMI to judge whether a trend exists and the SAR to manage the exit.

DMI vs. TDI

The Traders Dynamic Index (TDI) is a different indicator that shares only part of the name. It is built from the relative strength index (RSI) with moving averages and volatility bands around it, so it measures momentum and overbought or oversold conditions rather than directional movement. Using both means combining a momentum oscillator with a trend-direction tool, and they will not always agree.

What works well with the DMI

The DMI says little about where to enter or exit, so it is often combined with tools that do:

• ADX to filter out low-strength periods.
• Moving averages to define the larger trend, taking only DI crossovers in that direction.
• RSI or the stochastic oscillator to avoid entering after an extended move.
• Fibonacci retracement levels or support and resistance to place entries on pullbacks within a confirmed trend.
• Bollinger Bands or volume indicators to judge whether a breakout has participation behind it.

Whatever the combination, test it on historical data across several instruments and periods, including trading costs, before risking money. A rule that performs well on one chart and one year can fail elsewhere.

FAQs

The Bottom Line

The DMI suits traders who want to confirm the direction of a trend before acting on a breakout or pattern. Its crossovers lag and whipsaw in sideways markets, so pairing them with a rising ADX or a higher-timeframe trend filter removes many false signals. Because any signal can fail, size positions to cap the loss per trade and test the full rule set on historical data, including trading costs, before risking money.