RSI Divergence Indicator for MT4: The Settings That Tested Best

There is no divergence line in MetaTrader 4. You get RSI in the Oscillators folder, and everything after that — what counts as a swing point, how far apart the two points may be, when the pattern is confirmed — is a decision you make, usually without noticing you have made it. Those unnoticed decisions are worth more than the indicator settings people argue about. I measured each one across 16,926 real EURUSD 5-minute bars, June 3 to August 26, 2026, using the method from my RSI divergence backtest: entry on the confirmation bar, no look-ahead, pips before spread.

How do you set up RSI divergence on MT4?

The manual setup takes about a minute and costs nothing, and it is worth doing before you install anything third-party.

• Attach the oscillator: Insert → Indicators → Oscillators → Relative Strength Index. Period 14, apply to Close. Leave the 30 and 70 levels on — you will not trade them, but they mark the zones referenced below.
• Give it room: drag the indicator window taller. Divergence is a comparison between two peaks and a squashed subwindow hides small differences.
• Mark swing points, not wiggles: a low counts only if the five bars before and after it are higher. That is the single most important rule in this article and the one an eyeball scan usually breaks.
• Set a distance limit: ignore pairs of pivots more than 60 bars — five hours — apart. Beyond that you are connecting unrelated events.
• Wait for the fifth bar: the pattern is not tradable until the second swing point is confirmed. On this data that wait cost a median of 3.9 pips, and there is no version of the setup that avoids it.

If you prefer an indicator that draws the lines for you, the same rules apply — check what its swing-point setting is before you trust a single arrow, and check whether it redraws. Half the free divergence tools on the forums do, for the reasons I set out in the arrow repaint test.

Which RSI period and swing strength worked best?

Twelve combinations, all measured with a two-hour hold on the same bars.

Read the columns in the right order and the table says something simple. Every three-bar row is negative after costs, in all three RSI periods, because loose swing points manufacture patterns out of noise. Every five-bar row is positive. The RSI period itself barely moves the needle — 7, 14 and 21 land within 0.64 pips of each other once the pivots are strict.

The 12-bar rows look spectacular and I would not use them. Thirty-one trades at a 71% win rate across three months is the kind of number that appears and disappears with the sample; it is also, by construction, a signal you learn about an hour after the turn. If you want the honest version of that trade-off, it is the same one I ran into with parameter tuning in my optimised against default settings test — the best-looking row on historical data is rarely the one that repeats.

How far apart may the two pivots be?

The median gap between the two swing points that actually produced signals was 22 bars — one hour fifty — with a range of 6 to 60. Sixty bars is the practical cap: stretch to 120 and you gain 19 signals while the average trade drops. This is the setting nobody documents and it is worth more than the RSI period.

Should you only take divergence from the extremes?

This is the most common piece of divergence advice on the internet and the data cannot support it either way. Requiring RSI to be under 30 at a bullish divergence left three trades in three months. The 35/65 version left fourteen, and they were good, but fourteen is a handful. The honest reading is that the extremes are so rare at the exact moment a divergence confirms that the filter is not a setting — it is a decision to almost never trade.

Where do you put the stop?

Settings arguments are cheap; this table is where the money is. Same signals, same platform, an eight-hour maximum time in the trade, and a fixed stop and target in pips.

Identical entries, and the result swings from -0.24 to +5.31 pips per trade purely on where the exits sit. The 10-pip stop is the killer: a quarter of these trades go more than 10 pips underwater before working, so a tight stop converts winners into losers wholesale. The excursion numbers behind that are plain — the average trade’s worst moment is 6.5 pips against you and its best is 8.0 pips in your favour, which is why the stop has to be wider than the target is greedy. My stop distance test found the same 10-pip floor across a completely different entry rule, and the fixed against ATR stop comparison confirms it from the volatility side.

What hours should you watch for it?

The two negative rows are the thin hours at either end of the day, which is exactly where my best hours to trade EURUSD study put the poorest conditions for every rule I have tested. Sample sizes here are small — 13 to 22 trades a session — so treat this as a reason to skip the quiet hours rather than a reason to trade only London.

The settings I would actually run

• RSI 14, applied to Close — the default. Changing the period is the least productive thing you can do here.
• Five-bar swing points on both sides. Three bars turns the whole thing negative; twelve gives you a beautiful table and 31 trades.
• Maximum 60 bars between the two pivots, median gap around 22.
• 20-pip stop, 40-pip target, out after eight hours. +5.31 pips a trade on this data; a 10-pip stop was worth -0.24.
• Skip the Asian session and the late evening. Both were negative.
• No overbought filter. It leaves three trades a quarter.

Frequently asked questions

Does MT4 have an RSI divergence indicator built in?

No. MT4 includes RSI itself and nothing that draws divergence, so you either mark the swing points by hand or install a third-party indicator. Either way the swing-point rule is a choice you are making, and on this data it mattered more than the RSI period: three-bar swings lost money after spread while five-bar swings made it.

What is the best RSI period for divergence?

Practically, 14. Across 16,926 EURUSD 5-minute bars, RSI 21 was fractionally better at +1.71 pips per trade against +1.45 for RSI 14, and RSI 7 was worse at +1.07 — a spread small enough that the pivot definition dominates it.

Where should the stop go on a divergence trade?

Wider than instinct suggests. A 20-pip stop with a 40-pip target produced +5.31 pips per trade on these signals; a 10-pip stop with a 10-pip target produced -0.24. Around a quarter of the trades go over 10 pips against you before they work.

How many divergence signals should I expect?

About one every 13.6 hours on a single pair — 104 in nearly three months of EURUSD 5-minute data. It is not a signal you can trade all day on one chart.

Should I only trade divergence when RSI is oversold or overbought?

The data cannot recommend it. Requiring RSI below 30 or above 70 at the confirmation bar left three trades in the entire sample; a looser 35/65 requirement left fourteen. Both are too few to build a rule on.

Is divergence better than trading RSI levels?

Yes on this dataset — the only version of RSI that survived spread. The head-to-head, including hidden divergence and the 20/80 variant, is in my divergence against overbought comparison.

Does this work the same on MT5?

The RSI calculation is identical on both platforms, so the settings carry over unchanged; MT5 gives you more timeframes to look for pivots on, which is the practical difference I covered in my MT4 against MT5 comparison. If you are still setting the platform up, start with my first chart guide.