Multi Deviation Scaled Moving Average (Multi-DSMA)

1) What is the Multi-DSMA?

The Multi Deviation Scaled Moving Average (by ChartPrime) is an adaptive moving average that stacks eight Deviation-Scaled MAs (DSMAs) with progressively longer lookbacks. Each DSMA self-adjusts to current market activity via RMS scaling, then all eight are averaged into a single line. A score gauges how many fast layers are above the slowest layer, driving dynamic color/opacity and generating early/late crossover signals.

Use it when you need a trend-aware line that reacts quickly in acceleration phases but remains calm in choppy markets.

2) How it Works — The Short Version

Pipeline per bar:

1. Super Smoother prefilter → noise reduction with low lag.

2. RMS scaling (via exponential average) → dimensionless, volatility-aware signal.

3. Adaptive alpha → bigger |signal| ⇒ faster DSMA; clamped to avoid overshoot.

4. Eight DSMA layers → periods = Periods + k×step (k = 0..7).

5. Average of 8 layers → final “Multi DSMA Avg” line.

6. Score (0–1) → fraction of fast layers above the slowest layer.

7. Color/Opacity from score + Signals at 0.30/0.70 crossovers.

3) Core Mechanics

3.1 Super Smoother (per layer)

Each layer takes a zero-mean input (close - close[2]) and runs it through a 2-pole Super Smoother (Ehlers-style) using period-dependent coefficients. This removes high-frequency noise without the lag typical of simple MAs.

3.2 RMS Normalization

rms = sqrt(ExponentialAverage(period)(filt^2))
Dividing the filtered signal by its RMS (scaledFilt) makes the amplitude comparable across regimes (quiet vs. volatile), so the adaption depends on structure, not raw volatility.

3.3 Adaptive DSMA

alpha = min( |scaledFilt| * 5 / period , 1 )
DSMA = alpha * close + (1 - alpha) * DSMA[1]
When structure builds (|scaled| ↑), alpha increases and the DSMA hugs price more tightly. In calm conditions, alpha shrinks and the line smooths out.

3.4 The Eight Layers & Sensitivity Inversion

• Base period = Periods.

• Step = max(100 - Sensitivity, 1) → higher Sensitivity ⇒ smaller step ⇒ layers are closer together ⇒ more responsiveness.

• Layer periods: Periods, Periods+step, ..., Periods+7×step.

3.5 Multi-Layer Average

The plotted line “Multi DSMA Avg” is the simple average of the 8 DSMA values. This blend keeps the agility of fast layers while anchoring to the slower context.

3.6 Score Construction (0–1)

Compare each of the seven fast layers to the slowest (DSMA8). Each fast-above-slow adds ≈ 1/7.

• Score ≈ 1 → strong bullish stack (broad alignment).

• Score ≈ 0 → strong bearish stack.

• Middle → mixed/transition.

3.7 Dynamic Color & Opacity

• If score > 0.5: gradient fades in towards the upper color (bullish).

• If score ≤ 0.5: gradient fades out from the down color (bearish).
  Opacity follows the same logic: conviction (stack agreement) literally shows up as a more opaque line.

3.8 Signals (Hysteresis Band)

• Bullish signal when score crosses up 0.30 → early heads-up.

• Bearish signal when score crosses down 0.70 → late confirmation.
  This asymmetric band reduces flip-flops by asking for more evidence before flipping bearish.

4) How to Read It on the Chart

• Line direction: trend bias of the blended stack.

• Color & opacity: confidence. Deeper/denser = stronger alignment.

• Arrows:

  • ▲ at the Multi DSMA Avg when score crosses above 0.30.

  • ▼ at the Multi DSMA Avg when score crosses below 0.70.

• On-chart panel (last bar): two vertical bars show Up% vs Down% (score×100 vs 100–score×100), updated on the most recent update.

Tip: The score panel uses anchored drawing near the bottom-right of the chart. If you customize positions, keep the anchor(bottomright, xshift, yshift) logic consistent.

5) Typical Uses

A) Trend Confirmation / Bias Filter

• Trade only long when score > 0.5 and line slopes up; short when < 0.5 and slope down.

• Use arrows as timing aids, not stand-alone entries.

B) Momentum Entries

• Aggressive: act on the ▲ (0.30 cross up) if the line is already turning up and price pulls back to or slightly below the line.

• Conservative: wait for price to reclaim the line and the score to sustain > 0.6 for several bars.

C) Exits / Risk Control

• Partial exit when score slips below 0.7 from high levels (momentum fading).

• Hard exit on ▼ (0.70 cross down) and line turning flat/down.

D) Multi-Timeframe

• Use higher timeframe score as bias; execute with lower timeframe ▲/▼ around the Multi DSMA Avg.

This indicator is non-repainting (no future data; all calculations use current and past bars).

6) Inputs & Configuration

Periods (default 30)

• Sets the base lookback.

• Lower (e.g., 14–25): faster, more responsive, more signals.

• Higher (e.g., 40–80): smoother trend backbone, fewer signals.

Sensitivity (default 60, range 0–100)

• Controls step via step = max(100 - Sensitivity, 1).

• Higher Sensitivity ⇒ smaller step ⇒ layers cluster ⇒ more agility.

• Lower Sensitivity ⇒ larger step ⇒ wider spread ⇒ stronger “stack” requirement.

Recommended starting points:

• Scalping: Periods 20–30, Sensitivity 70–85

• Swing: Periods 30–50, Sensitivity 55–70

• Position: Periods 50–80, Sensitivity 40–60

Colors (RGB triplets)

• Upper color for bullish gradients, Down color for bearish.

• You can soften opacity extremes by lowering the alpha scaling, but the default provides a clear conviction cue.

7) ProBuilder Code

Share this