Is QQE a built-in MetaTrader 5 indicator?

No. MT5 has no iQQE, and no Hull average or Supertrend either. The Builder's node creates one iRSI handle and computes the smoothing, the band width and the trailing inside the generated EA. Nothing needs to be installed in the terminal and no custom .ex5 is involved.

What does the node actually output?

The trend, as +1 or −1 — not an RSI reading. Compare it against zero to use it as a filter, or cross it against a constant zero to catch only the moments it changes. The smoothed RSI, the bands and the trailing level are local to the generated code and are not published to other nodes.

It is the value standard implementations use, and it equals φ³ to five figures. Nothing in the mathematics requires it — it is a scale factor on the band, and it is the one setting that changes behaviour substantially. Measured on three synthetic series: 2.618 gave 74 to 81.5 flips per 1000 bars, 4.236 gave 53 to 59.5, and 6.854 gave 37.5 to 40.5.

How often does it signal?

About once every twenty bars before any filter — 53.0, 54.5 and 59.5 flips per 1000 bars on the three synthetic series measured here. The rate was nearly identical on a trending series and on a pure random walk, because the band adapts to the RSI's own volatility rather than to the market's direction. This is the main reason the template adds a gate instead of trading the flips directly.

Does the RSI-50 gate actually change anything?

Yes, and it was measured rather than assumed: it removed 33.6% to 39.6% of the flips across the three series. The gate asks a different question from the trend — the trend is built on a smoothed RSI and reports which side of its band that line is on, while the gate reads the level of the plain RSI — so it is not a restatement of the same condition.

Why does the node need a warm-up setting?

Because there is no QQE handle to copy from, the EA rebuilds the whole recursion on each tick from the last N bars, and the recursion has to run long enough to forget where it started. Measured against a full-history baseline, 200 bars reproduced the trend exactly on all three series; 100 bars differed on up to 0.8% of bars (two of the three series; the third matched exactly) and 50 bars on 5.6% to 8.2% of every series. The default of 300 sits comfortably past that.

My chart's QQE looks different from this one — why?

Implementations differ in the details, and one of them matters a great deal: whether the trend is judged against the band before or after it is updated on the current bar. Comparing against the updated band agreed with the correct version on only 47.2% to 48.4% of bars in the measurements here. Other differences are smaller — the two common ways of writing the trailing logic, with two bands or with a single level, produced identical trends on 100% of the 1500 bars compared.

Can I build this without coding?

Yes. The template is a QQE node and a constant zero into two Cross nodes for the flips, an RSI node with two Compare nodes against 50, two And gates, two Close nodes wired ahead of the entries, and Open Buy / Open Sell on a fixed lot. The RSI period, smoothing, band factor and warm-up are set in the Builder; the compiled EA exposes them as inputs along with the stop distances, lot size and maximum spread.

QQE for MT5 traders

What QQE tells you

QQE is a trend rule applied to momentum rather than to price. It smooths the RSI with a short EMA, measures how far that smoothed line travels from one bar to the next, smooths that too, and multiplies the result by 4.236 to get a band width. The band sits either side of the smoothed RSI and is allowed to move only one way — towards the line, never away from it — until the line breaks through, at which point the trend flips and the other band takes over. So the output is not an RSI level to compare against 70 or 30; it is a state, and the whole indicator is a way of asking whether momentum has changed side by more than its own recent volatility. That framing explains both the appeal and the problem. Because the band adapts, a quiet stretch tightens it and a volatile one widens it. And because the RSI itself moves constantly, the state changes often: measured on three synthetic series the trend flipped 53 to 60 times per 1000 bars.

Not an MT5 built-in — one iRSI handle, and everything else folded in the EA
The output is +1 / −1, not a 0–100 reading, so it is compared against zero
It flips roughly once every twenty bars, so it wants a filter rather than a bare entry

1 Smooth the RSI with an EMA of the smoothing length (5)
2 Smooth |change in that line| twice with a Wilder-length EMA, then scale by 4.236
3 Trail a band that width either side; the side broken is the trend

공식 및 계산 세부 정보 표시

MetaTrader 5 has no QQE, so there is no source file to quote. What the Builder generates is the common definition, with everything derived from one RSI handle:

ks = 2 / (Smooth + 1)                  <- 5  by default
kw = 2 / ((2*Period - 1) + 1)          <- 27 by default, so kw = 1/14

ma[i]   = rsi[i]*ks + ma[i-1]*(1-ks)                  <- smoothed RSI
atr[i]  = |ma[i] - ma[i-1]|*kw + atr[i-1]*(1-kw)      <- how far it moves, smoothed
datr[i] = atr[i]*kw + datr[i-1]*(1-kw)                <- smoothed again
dar     = datr[i] * Factor                            <- Factor = 4.236 = phi^3

The band is then trailed one way only, and the trend is read from which side the line breaks:

newLb = ma - dar,  newSb = ma + dar
lb = (ma[-1] > lb[-1] and ma > lb[-1]) ? max(lb[-1], newLb) : newLb
sb = (ma[-1] < sb[-1] and ma < sb[-1]) ? min(sb[-1], newSb) : newSb

crossedShort = (ma[-1] > sb[-1]) != (ma > sb[-1])     <- against the PREVIOUS band
crossedLong  = (ma[-1] > lb[-1]) != (ma > lb[-1])
trend = crossedShort ? +1 : crossedLong ? -1 : trend[-1]

The comparison uses sb[-1] and lb[-1] — the band as it stood before this bar’s update. Using the updated band instead compiles, runs, and draws a line that looks like QQE; measured against the correct version it agreed on 47.2% to 48.4% of bars across three synthetic series. A band that has already moved towards the line is much harder to break, so the flips it does report are a different set rather than a delayed one.

Because none of this comes from an indicator handle, the generated EA folds the recursion from scratch on every tick:

for k = warmup+1 down to shift:   <- oldest bar first, in series order
    ... the four lines above ...

That is what the warm-up setting buys. Against a baseline computed over the whole history, 200 bars reproduced the trend exactly on all three test series; 100 bars differed on up to 0.8% of bars — on two of the three, with the third matching exactly — and 50 bars on 5.6% to 8.2% of every series. The default of 300 is one comfortable step past where the error disappeared.

신호

QQE이 제공하는 백테스트 가능한 뚜렷한 신호 — 및 각 신호에 적합한 시장 상황.

Trend Flip

Builder로 구현 가능

조건: The QQE trend crosses zero — the smoothed RSI has broken the opposite band
최적 시장 상황: Trending
일반적인 사용: The template's entry, gated by the plain RSI's side of 50

Flip with a Momentum Gate

Builder로 구현 가능

조건: An upward flip while the raw RSI is above 50 (mirrored for shorts)
최적 시장 상황: Trending
일반적인 사용: The template as shipped. Measured, the gate removes a third of the flips

State as a Filter

Builder로 구현 가능

조건: QQE is +1 while another rule wants to buy
최적 시장 상황: Either regime
일반적인 사용: The cheapest honest use — no entry of its own, one Compare against zero

Divergence against the smoothed RSI

고급 로직

조건: Price makes a new extreme while the smoothed RSI does not
최적 시장 상황: Ranging
일반적인 사용: Widely described, but the smoothed RSI is not published by the node — only the trend is

MT5 구현

MetaTrader 5가 실제로 계산하고 그리는 내용 — 이 페이지의 모든 규칙이 기준으로 삼는 사양입니다.

버퍼

인덱스	버퍼	MT5 표시 방식	저장 값
0	MAIN_LINE	Line	There is no QQE buffer to read, because there is no QQE indicator in MT5. The only handle the generated EA creates is iRSI, and index 0 of that is the plain RSI everything else is built from. The smoothing, the band width and the trailing are computed in the EA itself, and none of the intermediate lines — smoothed RSI, band, trailing level — is published anywhere a node could read.

플랫폼 참고

MQL5 함수

iRSI

적용 가격

이 지표는 적용 가격 인수를 받지 않습니다.

Builder에서

QQE is not an MT5 function, so there is no signature to match. The applied price is fixed to Close, because that is what the standard implementations use and because the RSI here is a material rather than a reading — changing it silently changes every stage above it. The node exposes the RSI period, the RSI smoothing length, the band factor, a warm-up depth, shift, timeframe and a variable name.

구현 노트

★★★ MT5 has no iQQE. The node creates a single iRSI handle and folds everything else in OnTick: an EMA of the RSI at the smoothing length, an EMA of the absolute change in that line at the Wilder length 2 × period − 1, a second EMA of that at the same length, and then the band recursion. The Wilder length at the default resolves to a smoothing constant of 2/(2 × 14) = 1/14, which is why 27 and 14 are the same thing here written two ways.
★★★ The trend is judged against the band *before* it is updated on this bar. Getting that wrong is the single easiest way to produce a plausible but different indicator: compiled, it runs, and the line looks similar. Measured, comparing against the updated band instead agreed with the correct version on only 47.2% to 48.4% of bars across three synthetic series — worse than a coin toss, because a band that has already moved towards the line is much harder to break.
★★ The warm-up is a real number, not a guess. Because nothing is published by an indicator handle, the EA rebuilds the whole recursion on every tick from the last N bars. Against a baseline computed over the full history: 200 bars reproduced the trend exactly on all three series, 300 and 500 likewise; 100 bars differed on up to 0.8% of bars — on two of the three series, the third matching exactly — and 50 bars differed on 5.6% to 8.2% on all of them. The default is 300, which is one full doubling past the point where the error disappeared.
★★ Two implementations, one answer. QQE is usually written either with two bands and a trend flag, or with a single trailing level whose side is read directly. Both were folded here and compared over 500 bars on three synthetic series: the trend agreed on 100% of bars. The lines they *draw* are not the same — one publishes a band, the other a level — but the direction they call is, which is worth knowing before switching implementations because of a chart that looks different.
★★ The flip rate is the number to plan around. Measured over 2000 bars on each of three series: 53.0, 54.5 and 59.5 flips per 1000 bars. That is roughly one every twenty bars regardless of whether the series was trending or a pure random walk — the band adapts to the RSI's own volatility, so a quiet market does not produce a quiet indicator. Any use that trades every flip inherits that rate.
★ The node publishes only the trend. The smoothed RSI, the band and the trailing level exist as local variables in the generated code and are not available to any other node. A flow that wants the smoothed RSI has to build it, and the Builder has no EMA-of-an-indicator node — so in practice the trend is what you get.
★ The RSI's applied price is fixed to Close. Standard implementations do the same, and here it also protects the stages above it: the band width is measured from the smoothed RSI's own movement, so a different applied price would change the band as well as the line.

가장 잘 작동하는 경우 / 주의해서 사용

어떤 지표도 보편적인 우위를 가지지 않습니다. QQE이 도움이 되는 곳 — 그리고 오해를 일으키는 곳.

가장 잘 작동하는 경우

As a filter on a rule that already has a reason to trade
Trending instruments where a momentum regime lasts longer than the flip rate
Timeframes high enough that twenty bars is a meaningful stretch of time
Settings you have measured, because the factor changes the flip rate by a factor of two

주의해서 사용

Trading every flip — that is 53 to 60 entries per 1000 bars before any filter
Reading the output as an RSI level; it is a state, and the RSI itself is not published
Comparing implementations by eye — the bands differ even where the trend agrees
Lowering the factor to catch turns earlier; at 2.618 the flip rate rises by about half

QQE 전략 구축

신호를 진입 및 청산 규칙에 연결하고, 컴파일 가능한 MT5 EA를 내보내세요 — 코드 없이.

One QQE node at 14 / 5 / 4.236, reading the last closed bar
A constant of zero, crossed against the trend to catch only the flips
One RSI node at 14, compared against 50 for the gate
Flip up and RSI above 50 → Close short, then Open Buy · SL 60 / TP 180
Flip down and RSI below 50 → Close long, then Open Sell, mirrored

QQE을 다른 지표와 결합

하나의 지표만으로는 충분하지 않습니다. 이 조합들이 QQE의 취약점을 보완합니다.

QQE + RSI

QQE trend crossing zero
RSI against 50
And gate

이유: The gate the template ships with, and the pairing is less redundant than it looks: the QQE trend is built from a *smoothed* RSI and reports only which side of its band that line is on, while the raw RSI still carries the level. Asking for both means asking that momentum has turned and that it is on the side of 50 the turn claims. Measured on three synthetic series, the gate removed 33.6% to 39.6% of the flips — so it is doing real work rather than restating the trend
최적 시장 상황: Trending

이 전략 구축 →

QQE + Supertrend

QQE trend
Supertrend trend
And gate

이유: The same trailing-band idea applied to price instead of to momentum. Reading the two pages together is the cheapest way to see what the choice of space costs: Supertrend's band is a multiple of ATR around the midpoint of a bar, so it cannot flip while price is still, whereas QQE's band is a multiple of the RSI's own movement and can flip in a range where price has gone nowhere. Requiring both to agree is the usual way to use them together
최적 시장 상황: Trending

No template holds both. Add a Supertrend node beside the QQE node, compare each against zero, and join them with an And gate.

Builder 열기 →

QQE + ADX

ADX above 25
QQE trend crossing zero
And gate

이유: The measured problem with this indicator is the flip rate — about one every twenty bars on synthetic data, and the band adapts so that a quiet stretch produces just as many. ADX answers a question QQE cannot ask, because nothing in QQE looks at price at all: is there a trend in the instrument, rather than in its momentum. It removes the regime where the extra sensitivity is a liability instead of an edge
최적 시장 상황: Trending

Add an ADX node with a Compare against 25 and join it to the Cross with an And gate before the entry.

Builder 열기 →

매개변수

자신의 통화쌍과 시간 프레임에서 검증할 초기값 — 보장된 설정이 아닙니다.

매개변수	기본값	권장 테스트 범위	기능
RSI period	14	2–999	The length of the RSI everything is built on, and it also sets the smoothing length of the two band stages: those use a Wilder length of 2 × period − 1, which is 27 at the default. So raising this makes the underlying momentum slower and the band steadier at the same time — the two cannot be separated from the Builder.
RSI smoothing	5	1–200	The EMA applied to the RSI before anything else. It is what makes the line smooth enough for a band to be meaningful; at 1 the smoothing disappears and the band is measured on the raw RSI's bar-to-bar noise. This is the setting to raise if the flips are too frequent and you want to keep the factor at its standard value.
Band factor	4.236	0.1–100	How many times the smoothed movement the band sits away from the line. 4.236 is the value standard implementations use — it is φ³ to five figures — and it is the one real speed dial. Measured on three synthetic series, 2.618 produced 74 to 81.5 flips per 1000 bars, 4.236 produced 53 to 59.5, and 6.854 produced 37.5 to 40.5. Halving the flip rate takes a factor of roughly 1.6, not a small tweak.
Warm-up bars	300	50–5000	How many bars of the recursion the generated code folds on each tick. Nothing here is published by an indicator, so the EA rebuilds the smoothing and the band from scratch every time. Measured against a baseline computed over the full history, 200 bars reproduced the trend exactly on all three test series; 100 bars differed on up to 0.8% of bars — two of the three series, with the third matching exactly — and 50 bars differed on 5.6% to 8.2% on every series. The default of 300 is the safe side of that measurement.
Shift	1	0–100	Which bar the node reports. The template uses 1 so decisions are made on a closed bar. Shift 0 is worth avoiding here in particular: the band is trailing, so an unfinished bar can flip the state and then unflip it before the bar closes.
Timeframe	Current	Current / M1–MN1	Which timeframe the RSI handle is created on. The flip rate quoted on this page is per bar, so a rate of once every twenty bars means twenty minutes on M1 and eighty hours on H4 — the same indicator is a scalping tool or a position filter depending only on this.

초기 프리셋

Template 14 / 5 / 4.236 with an RSI-50 gate The standard settings, with the gate that removes about a third of the flips

Slower 14 / 5 / 6.854 Measured at 37.5–40.5 flips per 1000 bars against the default's 53–59.5

Filter only Compare the trend against zero No entry of its own — one node and one Compare, and nothing to fit

시장 예시

QQE이 작동하는 곳, 실패하는 곳, 그리고 필터가 결과를 어떻게 바꾸는지.

성공

The regime change caught on the flip

Momentum turns and the smoothed RSI breaks the band it had been trailing above. The plain RSI is already through 50, so the gate lets the entry through, and the state stays on that side for the length of the move.

실패

The flip in a range

Price has gone nowhere, but the RSI has been moving inside it, and the band has tightened around a quiet stretch of momentum. The break happens on a move that means nothing on the price chart — the case the RSI-50 gate and an ADX filter are both aimed at.

필터링됨

The flip that the gate refused

An upward flip while the raw RSI is still below 50. Measured on synthetic series a third of all flips look like this, which is the single largest change the template makes to the bare indicator.

FAQ

Is QQE a built-in MetaTrader 5 indicator?: No. MT5 has no iQQE, and no Hull average or Supertrend either. The Builder's node creates one iRSI handle and computes the smoothing, the band width and the trailing inside the generated EA. Nothing needs to be installed in the terminal and no custom .ex5 is involved.
What does the node actually output?: The trend, as +1 or −1 — not an RSI reading. Compare it against zero to use it as a filter, or cross it against a constant zero to catch only the moments it changes. The smoothed RSI, the bands and the trailing level are local to the generated code and are not published to other nodes.
Why 4.236?: It is the value standard implementations use, and it equals φ³ to five figures. Nothing in the mathematics requires it — it is a scale factor on the band, and it is the one setting that changes behaviour substantially. Measured on three synthetic series: 2.618 gave 74 to 81.5 flips per 1000 bars, 4.236 gave 53 to 59.5, and 6.854 gave 37.5 to 40.5.
How often does it signal?: About once every twenty bars before any filter — 53.0, 54.5 and 59.5 flips per 1000 bars on the three synthetic series measured here. The rate was nearly identical on a trending series and on a pure random walk, because the band adapts to the RSI's own volatility rather than to the market's direction. This is the main reason the template adds a gate instead of trading the flips directly.
Does the RSI-50 gate actually change anything?: Yes, and it was measured rather than assumed: it removed 33.6% to 39.6% of the flips across the three series. The gate asks a different question from the trend — the trend is built on a smoothed RSI and reports which side of its band that line is on, while the gate reads the level of the plain RSI — so it is not a restatement of the same condition.
Why does the node need a warm-up setting?: Because there is no QQE handle to copy from, the EA rebuilds the whole recursion on each tick from the last N bars, and the recursion has to run long enough to forget where it started. Measured against a full-history baseline, 200 bars reproduced the trend exactly on all three series; 100 bars differed on up to 0.8% of bars (two of the three series; the third matched exactly) and 50 bars on 5.6% to 8.2% of every series. The default of 300 sits comfortably past that.
My chart's QQE looks different from this one — why?: Implementations differ in the details, and one of them matters a great deal: whether the trend is judged against the band before or after it is updated on the current bar. Comparing against the updated band agreed with the correct version on only 47.2% to 48.4% of bars in the measurements here. Other differences are smaller — the two common ways of writing the trailing logic, with two bands or with a single level, produced identical trends on 100% of the 1500 bars compared.
Can I build this without coding?: Yes. The template is a QQE node and a constant zero into two Cross nodes for the flips, an RSI node with two Compare nodes against 50, two And gates, two Close nodes wired ahead of the entries, and Open Buy / Open Sell on a fixed lot. The RSI period, smoothing, band factor and warm-up are set in the Builder; the compiled EA exposes them as inputs along with the stop distances, lot size and maximum spread.

Glossary

핵심 용어

백테스트 과최적화 스프레드

QQE

What QQE tells you

신호

Trend Flip

Flip with a Momentum Gate

State as a Filter

Divergence against the smoothed RSI

MT5 구현

버퍼

플랫폼 참고

가장 잘 작동하는 경우 / 주의해서 사용

가장 잘 작동하는 경우

주의해서 사용

QQE 전략 구축

QQE을 다른 지표와 결합

QQE + RSI

QQE + Supertrend

QQE + ADX

매개변수

초기 프리셋

시장 예시

The regime change caught on the flip

The flip in a range

The flip that the gate refused

관련 지표

FAQ

관련 글

핵심 용어