Is VLDMI built into MetaTrader 5?

No, and there is no single function to call. The Builder assembles it from standard parts: one iATR over the short period, one over twice the base period, and one iRSI for every candidate period the settings allow. At the defaults that is thirteen handles created by a single node, which is more than any other node in the palette.

How does the period actually change?

The node computes (1 − shortATR / longATR) × steps, rounds it to a whole number and adds it to the base period, then reads whichever pre-built RSI handle is closest to that. Because the short ATR is in the numerator, a burst of volatility makes the ratio larger than one, the expression negative, and the period shorter.

How often does it really adapt?

That depends on the market, not the indicator, and the range is wide. With volatility held constant the chosen period was the base 14 on 93.5% of bars. With volatility regimes 200–500 bars long it left the base on 18.7% of bars, and with regimes only 20 bars long — shorter than the ATR window itself — on 65.9%. Both ATR windows overlap heavily, so the ratio only swings when volatility changes faster than they can follow.

Should I raise the step count to make it more adaptive?

The measurements argue against it. At the default five steps the candidate list runs 9 to 19, and across every regime tested the periods actually selected were 10 to 17 — with the two most violent regimes reaching 18. Nine of the eleven handles were ever read, and the target period landed outside the list on 0.00% of bars. Raising the step count builds more handles at the ends of a list that the ratio does not reach; shortening the short ATR period widens the ratio's swing, which is the input that actually controls this.

Is it better than a plain RSI?

It is different on a minority of bars and never opposite. Against a fixed RSI(14) with the same thresholds and slope condition, the signals differ on 4.8% of signal bars in slow-changing volatility and 14.0% in the fastest regime, with the adaptive version mostly adding signals rather than removing them — and across roughly 400,000 evaluations neither ever fired against the other. Whether the added signals are worth having is a question about your instrument and your costs, not one a synthetic series can answer, so this page does not claim it.

Why does the template close on time instead of on the indicator?

Because the signal is a one-bar pulse and there is nothing to wait for. Measured after a signal, the RSI is out of the threshold zone in a median of 2 bars, so an exit that waits for the condition to clear would close almost immediately. Waiting for the reading to return to the midline takes a median of 19 bars — mean 23.2, 75th percentile 31 — so the Time Exit is set at 20 bars held. At that point 98.9% of cases have left the zone and 55.2% have crossed 50.

Why is the stop the same distance as the target?

Because the measured excursions are nearly the same size. Normalising by the ATR at entry, the worst adverse move over the twenty bars had a median of 1.93 × ATR and the best favourable move 1.79 × ATR. The 1 : 3 stop-to-target shape the trend templates on this site use would point the wrong way here: a target three ATRs away is reached 25.8% of the time while a stop three ATRs away is hit 29.1% of the time. The distances are a guard; time is the exit.

Why does it not take every signal?

The template holds one position at a time and holds it for twenty bars, and 60.7% of signals arrive within twenty bars of the previous one. That is a deliberate trade-off rather than an oversight: taking them all means stacking positions in the same direction during exactly the stretches where a fade is most likely to be wrong. If you want them, remove the Time Exit and raise the position limit — and re-test, because that is a materially different strategy.

Can I build this without coding?

Yes. The VLDMI template has one VLDMI node at the defaults feeding two Compare nodes against zero, which open a buy or a sell with 70 / 70 pip distances, plus a Time Exit closing anything held twenty bars. The base period, the ATR period, the step count, the levels, the slope threshold and its lookback, the stops and the exit length are all EA inputs.

VLDMI for MT5 traders

What VLDMI tells you

Most adaptive indicators change a weight. This one changes a period: it measures how a 10-bar ATR compares with a 28-bar one, turns that ratio into a whole number of steps, and reads the RSI of the period that many steps away from 14. Because MetaTrader cannot change an indicator's period after a handle exists, the Builder solves it by building every period the settings could ask for — eleven of them at the defaults — and switching between them bar by bar. What comes out is not a line but one of three integers, so the flow reads it with a comparison against zero. The interesting part is what the measurements say about the switching: how much it moves depends entirely on how fast volatility changes, and the extremes of the list it so carefully builds are never used.

One node, thirteen handles — and the count is decided at runtime by the inputs
The output is a one-bar pulse of +1 / -1 / 0, read with Compare against zero
The adaptation gates the same rule; it never reverses the direction of a signal

1 Compare a 10-bar ATR with a 28-bar ATR and scale the difference by the step count
2 Round to a whole number of steps and pick the nearest pre-built RSI period
3 Report +1 if that RSI is oversold and falling fast, -1 if overbought and rising fast

공식 및 계산 세부 정보 표시

MetaTrader cannot change an indicator’s period once its handle exists. A variable-period RSI therefore cannot be one indicator — it has to be a shelf of them, with something choosing which to read.

short = ATR(10)     long = ATR(28)          <- 28 = base x 2
x     = (1 - short / long) * steps          <- steps = 5
period = 14 + round_half_to_even(x)         <- not MathRound
read the nearest pre-built RSI handle to that period

At the defaults the shelf holds periods 9 to 19 — eleven iRSI handles — plus the two iATR. Thirteen handles from one node, and the count is not fixed in the code: it is computed in OnInit from the inputs, into a dynamically sized array. No other node in the palette does that.

Three of the thirteen are never read. Across five volatility regimes — from perfectly steady to a fresh regime every ten bars with a hundredfold range — the periods actually selected were 10 through 17, reaching 18 only in the two most violent cases:

Volatility regime	Bars off the base period	Candidate periods used
Constant	6.5%	13, 14, 15 (3 of 11)
Changes every 200–500 bars	18.7%	10–17 (8 of 11)
Changes every 20 bars	65.9%	10–18 (9 of 11)
Changes every 10 bars	63.6%	10–18 (9 of 11)

The target period landed outside the candidate range on 0.00% of bars in every regime, so the clamping the search performs never fires either. Both ATR windows overlap heavily, so the ratio simply does not swing far enough to reach the ends of the list the node builds for it.

How adaptive it is, is a fact about the market. The same settings gave 6.5% and 65.9% on the two extremes above. Any single number for this is a number about a data set.

It gates the same rule rather than replacing it. Run a fixed RSI(14) with identical thresholds and slope condition: the two disagree on 4.8% of signal bars in slow-changing volatility and 14.0% in the fastest regime, and across roughly 400,000 evaluations neither ever fired in the opposite direction to the other.

신호

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

Oversold and Still Falling

Builder로 구현 가능

조건: The chosen RSI is below 30 and has dropped more than 10 points over five bars
최적 시장 상황: Ranging
일반적인 사용: The template's long entry — both halves are required, a low reading alone does nothing

Overbought and Still Rising

Builder로 구현 가능

조건: The chosen RSI is above 70 and has risen more than 10 points over five bars
최적 시장 상황: Ranging
일반적인 사용: The mirror image, and the reason the output is signed rather than boolean

The Period Sits Still

Builder로 구현 가능

조건: The ATR ratio stays near one and the chosen period stays on the base
최적 시장 상황: Either regime
일반적인 사용: Not a trade — the state in which this is a fixed RSI wearing an adaptive name

Low Reading Without the Drop

고급 로직

조건: The RSI is under 30 but has moved less than the slope threshold
최적 시장 상황: Ranging
일반적인 사용: Deliberately not a signal — requires lowering the slope, which is a different setup

MT5 구현

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

플랫폼 참고

MQL5 함수

iATR + iRSI (no single MT5 function)

적용 가격

이 지표는 적용 가격 인수를 받지 않습니다.

Builder에서

There is no indicator signature to match. The RSI handles are built with PRICE_CLOSE hard-coded in the codegen and the ATR handles take no applied price at all, so the node exposes none. What it does expose is the base RSI period, the short ATR period, the number of shift steps, the oversold and overbought levels, the minimum RSI change, the lookback that change is measured over, a shift and a timeframe.

구현 노트

There is no MT5 function for this. The node is assembled in the generated EA from iATR(short), iATR(base × 2) and one iRSI per candidate period — at the defaults, 11 + 2 = 13 handles from a single palette node. The RSI handles live in a dynamically sized array because the count is not known until the inputs are read, which makes this the only node in the palette whose handle count is decided at runtime.
★ The shift is computed as (1 − shortATR / longATR) × steps and rounded to a whole number. The rounding is banker's rounding — halves go to the even side — because the reference implementation this was ported from uses Python's round() rather than MQL's MathRound, and the two disagree on exact halves. Measured over roughly 400,000 evaluations across five volatility regimes, an exact half never occurred once, so in practice the distinction is invisible; the codegen reproduces it anyway so that the EA and the reference agree by construction rather than by luck.
★ The nearest-period search walks the ascending candidate list with a strict less-than, so a tie resolves to the smaller period. This matters because the candidate list is symmetric around the base: a target exactly between two candidates is not a rounding artefact but a routine outcome of an integer target landing on a boundary.
★★ The extremes of the candidate list are never selected. At the defaults the list is 9…19, and across five volatility regimes — from constant volatility to a fresh regime every ten bars with a hundredfold range — the periods actually chosen were 10 through 17, or 10 through 18 in the two most violent cases. Nine of eleven handles were ever read. The target period landed outside the candidate range on 0.00% of bars in every regime, so the clamping the search performs never fires either.
★ How much the period moves is a property of the market, not of the indicator. With volatility held constant the chosen period leaves the base on 6.5% of bars; with regimes 200–500 bars long, 18.7%; with regimes of 20 bars, 65.9%. Any statement about 'how adaptive VLDMI is' without stating the regime is meaningless, which is why this page gives the range rather than a single number.
The node evaluates the whole selection twice per tick — once for the bar you point at and once for the bar before it — so that a Cross node can be attached to the output. The template uses Compare, which reads only the first, so half of the CopyBuffer work is computed and thrown away on every tick.

가장 잘 작동하는 경우 / 주의해서 사용

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

가장 잘 작동하는 경우

Instruments whose volatility genuinely shifts within the ATR windows, not between weeks
M15 and above, where thirteen handles per node is a cost you only pay once
Ranging conditions — every signal it produces is a fade
Flows that give the position a defined lifetime, because the pulse has no exit of its own

주의해서 사용

Assuming the period moves: with steady volatility it stays on the base 93.5% of the time
Large step counts, which build handles for periods that are never selected
Reading the output as a level or a strength — it is +1, -1 or 0 and nothing else
Trending markets, where fading an overbought reading is the losing side by construction

VLDMI 전략 구축

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

One VLDMI node: base period 14, a ten-bar volatility window, five shift steps, 30 / 70 levels
Signal greater than zero → Open Buy · SL 70 / TP 70
Signal less than zero → Open Sell, same distances mirrored
A Time Exit closing anything held twenty bars or more
Shift 1, so the pulse is judged on a closed bar

VLDMI을 다른 지표와 결합

하나의 지표만으로는 충분하지 않습니다. 이 조합들이 VLDMI의 취약점을 보완합니다.

VLDMI + ATR

ATR at 14
Stop loss in ATR mode
Signal greater than zero

이유: The 70 / 70 distances are in pips, which travels badly between instruments and between quiet and violent weeks — and this setup is unusually exposed to that, because the measured excursions scale with volatility while the stop does not. Sizing both from a measured range makes the symmetry the page argues for hold everywhere rather than only on the instrument the numbers came from. It is also the honest way to express the finding that the adverse and favourable excursions are near-equal
최적 시장 상황: Either regime

The Stop Loss node has an ATR mode, so replacing the fixed distance needs no extra condition — swap the mode and set the multiplier.

Builder 열기 →

VLDMI + ADX

ADX below 25
Signal greater than zero
Buy

이유: Every signal this node produces is a fade, and the failure mode of a fade is a trend that keeps going. The node cannot see that: its own volatility measurement decides the RSI period, not whether the market is directional, and a strong trend raises volatility in a way that shortens the period and makes the fade fire sooner. A trend-strength reading is the missing input rather than a refinement of an existing one
최적 시장 상황: Ranging

No single-flow template combines them. Add an ADX node with a Compare below 25 and join it to the VLDMI Compare with an And gate.

Builder 열기 →

VLDMI + Bollinger Bands

Bollinger Bands at 20
Signal greater than zero
Buy

이유: A second opinion on the same event from a different measurement. The node's rule is entirely about the RSI's level and its slope; a band touch is about distance from a moving average in units of dispersion. When both agree the drop is both momentum-extreme and unusually far from the mean, and when only the RSI agrees you are fading a move that was never statistically large — which on a one-position template is the difference between spending the slot and keeping it
최적 시장 상황: Ranging

Add a Bollinger Bands node with a Compare of price against the lower band, then join it to the VLDMI Compare with an And gate.

Builder 열기 →

매개변수

자신의 통화쌍과 시간 프레임에서 검증할 초기값 — 보장된 설정이 아닙니다.

매개변수	기본값	권장 테스트 범위	기능
Base RSI period	14	2–1000	The period the shift is measured from, and also what the long ATR is built from — the long ATR period is this value doubled, so raising the base widens the volatility comparison as well as the RSI. Measured with steady volatility, the chosen period is this value on 93.5% of bars.
Short ATR period	10	2–1000	The recent-volatility half of the ratio. What matters is the gap between this and the base doubled: the closer the two windows, the more they overlap and the smaller the ratio's excursions, so shortening this is the direct way to make the period move more often.
Shift steps	5	0–50	How many periods either side of the base are available, and therefore how many iRSI handles get built: the candidate list runs from max(2, base − steps) to max(2, base + steps). At 5 that is eleven handles; at 20 it is thirty-three and at 50 it is sixty-three. The measured selections never reached the ends of the list, so a large value here buys handles that are created, warmed and never read.
Oversold level	30	0–100	The RSI level below which a long is allowed. Note this is only half of the condition — the RSI also has to have fallen by more than the slope threshold over the lookback, so a market that drifts down to 29 and sits there produces nothing.
Overbought level	70	0–100	The mirror. Both levels are read on whichever RSI period was selected for that bar, which is worth remembering when comparing readings against a chart running a fixed period.
Minimum RSI change	10	0–100	How many RSI points the reading must have moved over the lookback, in the direction being faded. This is what makes the output a pulse rather than a state: it is true on the bars where the move is happening, not for as long as the reading stays extreme.
Slope lookback	5	1–500	How many bars back the change is measured over. The node reads lookback + 1 values from the selected RSI handle each evaluation, so raising this raises the copy size rather than the handle count.
Shift	1	0–100	Which bar the node reads. The template uses 1, so decisions are made on closed bars. The node always evaluates two consecutive bars — the one you point at and the one before it — so that a Cross node could be attached; with a Compare node, as in the template, the second evaluation is computed and discarded.
Timeframe	Current	Current / M1–MN1	Which timeframe's bars all thirteen handles are built on. They all share it — there is no way to read the volatility ratio from one timeframe and the RSI from another.

초기 프리셋

Template 14 / 10 / 5 steps Eleven RSI handles, of which nine are ever selected

Fixed in disguise 14 / 10 / 0 steps One handle and no adaptation — worth running as the control before tuning anything

Faster switching 14 / 5 / 5 steps A shorter recent window overlaps the long one less, so the ratio swings further

시장 예시

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

성공

A drop faded near the low

Volatility rises, the chosen period shortens, the shorter RSI reaches oversold sooner than a fixed one would, and the pulse fires while the drop is still moving. The position is closed twenty bars later whether or not the reading has recovered.

실패

The trend that kept going

The same signal in a directional market. Nothing in the node measures direction — rising volatility shortens the period and makes the fade fire earlier, which is exactly the wrong way round when the move is real.

필터링됨

Low but not falling

The RSI sits below thirty for a stretch without moving much. No signal is produced, because the rule needs the reading to be extreme and still moving; a slow grind into oversold is not what it fades.

FAQ

Is VLDMI built into MetaTrader 5?: No, and there is no single function to call. The Builder assembles it from standard parts: one iATR over the short period, one over twice the base period, and one iRSI for every candidate period the settings allow. At the defaults that is thirteen handles created by a single node, which is more than any other node in the palette.
How does the period actually change?: The node computes (1 − shortATR / longATR) × steps, rounds it to a whole number and adds it to the base period, then reads whichever pre-built RSI handle is closest to that. Because the short ATR is in the numerator, a burst of volatility makes the ratio larger than one, the expression negative, and the period shorter.
How often does it really adapt?: That depends on the market, not the indicator, and the range is wide. With volatility held constant the chosen period was the base 14 on 93.5% of bars. With volatility regimes 200–500 bars long it left the base on 18.7% of bars, and with regimes only 20 bars long — shorter than the ATR window itself — on 65.9%. Both ATR windows overlap heavily, so the ratio only swings when volatility changes faster than they can follow.
Should I raise the step count to make it more adaptive?: The measurements argue against it. At the default five steps the candidate list runs 9 to 19, and across every regime tested the periods actually selected were 10 to 17 — with the two most violent regimes reaching 18. Nine of the eleven handles were ever read, and the target period landed outside the list on 0.00% of bars. Raising the step count builds more handles at the ends of a list that the ratio does not reach; shortening the short ATR period widens the ratio's swing, which is the input that actually controls this.
Is it better than a plain RSI?: It is different on a minority of bars and never opposite. Against a fixed RSI(14) with the same thresholds and slope condition, the signals differ on 4.8% of signal bars in slow-changing volatility and 14.0% in the fastest regime, with the adaptive version mostly adding signals rather than removing them — and across roughly 400,000 evaluations neither ever fired against the other. Whether the added signals are worth having is a question about your instrument and your costs, not one a synthetic series can answer, so this page does not claim it.
Why does the template close on time instead of on the indicator?: Because the signal is a one-bar pulse and there is nothing to wait for. Measured after a signal, the RSI is out of the threshold zone in a median of 2 bars, so an exit that waits for the condition to clear would close almost immediately. Waiting for the reading to return to the midline takes a median of 19 bars — mean 23.2, 75th percentile 31 — so the Time Exit is set at 20 bars held. At that point 98.9% of cases have left the zone and 55.2% have crossed 50.
Why is the stop the same distance as the target?: Because the measured excursions are nearly the same size. Normalising by the ATR at entry, the worst adverse move over the twenty bars had a median of 1.93 × ATR and the best favourable move 1.79 × ATR. The 1 : 3 stop-to-target shape the trend templates on this site use would point the wrong way here: a target three ATRs away is reached 25.8% of the time while a stop three ATRs away is hit 29.1% of the time. The distances are a guard; time is the exit.
Why does it not take every signal?: The template holds one position at a time and holds it for twenty bars, and 60.7% of signals arrive within twenty bars of the previous one. That is a deliberate trade-off rather than an oversight: taking them all means stacking positions in the same direction during exactly the stretches where a fade is most likely to be wrong. If you want them, remove the Time Exit and raise the position limit — and re-test, because that is a materially different strategy.
Can I build this without coding?: Yes. The VLDMI template has one VLDMI node at the defaults feeding two Compare nodes against zero, which open a buy or a sell with 70 / 70 pip distances, plus a Time Exit closing anything held twenty bars. The base period, the ATR period, the step count, the levels, the slope threshold and its lookback, the stops and the exit length are all EA inputs.

Glossary

핵심 용어

백테스트 과최적화 최대 낙폭

VLDMI

What VLDMI tells you

신호

Oversold and Still Falling

Overbought and Still Rising

The Period Sits Still

Low Reading Without the Drop

MT5 구현

플랫폼 참고

가장 잘 작동하는 경우 / 주의해서 사용

가장 잘 작동하는 경우

주의해서 사용

VLDMI 전략 구축

VLDMI을 다른 지표와 결합

VLDMI + ATR

VLDMI + ADX

VLDMI + Bollinger Bands

매개변수

초기 프리셋

시장 예시

A drop faded near the low

The trend that kept going

Low but not falling

관련 지표

FAQ

관련 글

핵심 용어