Linh's Anomaly Radar v2

What this script does

It’s an event detector for price/volume anomalies that often precede or confirm moves.
It watches a bunch of patterns (Wyckoff tests, squeezes, failed breakouts, turnover bursts, etc.), applies robust z-scores, optional trend filters, cooldowns (to avoid spam), and then fires:

A shape/label on the bar,

A row in the mini panel (top-right),

A ready-made alertcondition you can hook into.

How to add & set up (TradingView)

Paste the script → Save → Add to chart on Daily first (works on any TF).

Open Settings → Inputs:

General
• Use Robust Z (MAD): more outlier-resistant; keep on.
• Z Lookback: 60 bars is ~3 months; bump to 120 for slower regimes.
• Cooldown: min bars to wait before the same signal can fire again (default 5).
• Use trend filter: if on, “bullish” signals only fire above SMA(tfLen), “bearish” below.

Thresholds: fine-tune sensitivity (defaults are sane).

To create alerts: Right-click chart → Add alert

Condition: Linh’s Anomaly Radar v2 → choose a specific signal or Composite (Σ).

Options: “Once per bar close” (recommended).

Customize message if you want ticker/timeframe in your phone push.

The mini panel (top-right)

Signal column: short code (see cheat sheet below).

Fired column: a dot “•” means that on the latest bar this signal fired.

Score (right column): total count of signals that fired this bar.

Σ≥N shows your composite threshold (how many must fire to trigger the “Composite” alert).

Shapes & codes (what’s what)
Code Name (category) What it’s looking for Why it matters
STL Stealth Volume z(volume)>5 & ** z(return)
EVR Effort vs Result squeeze z(vol)>3 & z(TR)<−0.5 Heavy effort, tiny spread → absorption
TGV Tight+Heavy (HL/ATR)<0.6 & z(vol)>3 Tight bar + heavy tape → pro activity
CLS Accumulation cluster ≥3 of last 5 bars: up, vol↑, close near high Classic accumulation footprint
GAP Open drive failure Big gap not filled (≥80%) & vol↑ One-sided open stalls → fade risk
BB↑ BB squeeze breakout Squeeze (z(BBWidth)<−1.3) → close > upperBB & vol↑ Regime shift with confirmation
ER↑ Effort→Result inversion Down day on vol then next bar > prior high Demand overwhelms supply
OBV OBV divergence OBV slope up & ** z(ret20)
WER Wide Effort, Opposite Result z(vol)>3, close<open, z(CLVM)>+1 Selling into strength / distribution
NS No-Supply (Wyckoff) Down bar, HL<0.6·ATR, vol << avg Sellers absent into weakness
ND No-Demand (Wyckoff) Up bar, HL<0.6·ATR, vol << avg Buyers absent into strength
VAC Liquidity Vacuum z(vol)<−1.5 & ** z(ret)
UTD UTAD (failed breakout) Breaks swing-high, closes back below, vol↑ Stop-run, reversal risk
SPR Spring (failed breakdown) Breaks swing-low, closes back above, vol↑ Bear trap, reversal risk
PIV Pocket Pivot Up bar; vol > max down-vol in lookback Quiet base → sudden demand
NR7 Narrow Range 7 + Vol HL is 7-bar low & z(vol)>2 Coiled spring with participation
52W 52-wk breakout quality New 52-wk close high + squeeze + vol↑ High-quality breakouts
VvK Vol-of-Vol kink z(ATR20,200)>0.5 & z(ATR5,60)<0 Long-vol wakes up, short-vol compresses
TAC Turnover acceleration SMA3 vol / SMA20 vol > 1.8 & muted return Participation surging before move
RBd RSI Bullish div Price LL, RSI HL, vol z>1 Exhaustion of sellers
RS↑ RSI Bearish div Price HH, RSI LH, vol z>1 Exhaustion of buyers
Σ Composite Count of all fired signals ≥ threshold High-conviction bar

Placement:

Triangles up (below bar) → bullish-leaning events.

Triangles down (above bar) → bearish-leaning events.

Circles → neutral context (VAC, VvK, Composite).

Key inputs (quick reference)
General

Use Robust Z (MAD): keep on for noisy tickers.

Z Lookback (lenZ): 60 default; 120 if you want fewer alerts.

Trend filter: when on, bullish signals require close > SMA(tfLen), bearish require <.

Cooldown: prevents repeated firing of the same signal within N bars.

Phase-1 thresholds (core)

Stealth: vol z > 5, |ret z| < 1.

EVR: vol z > 3, TR z < −0.5.

Tight+Heavy: (HL/ATR) < 0.6, vol z > 3.

Cluster: window=5, min=3 strong bars.

GapFail: gap/ATR ≥1.5, fill <80%, vol z > 2.

BB Squeeze: z(BBWidth)<−1.3 then breakout with vol z > 2.

Eff→Res Up: prev bar heavy down → current bar > prior high.

OBV Div: OBV uptrend + |z(ret20)|<0.3.

Phase-2 thresholds (extras)

WER: vol z > 3, close<open, z(CLVM)>1.

No-Supply/No-Demand: tight bar & very light volume vs SMA20.

Vacuum: vol z < −1.5, |ret z|>1.5.

UTAD/Spring: swing lookback N (default 20), vol z > 2.

Pocket Pivot: lookback for prior down-vol max (default 10).

NR7: 7-bar narrowest range + vol z > 2.

52W Quality: new 52-wk high + squeeze + vol z > 2.

VoV Kink: z(ATR20,200)>0.5 AND z(ATR5,60)<0.

Turnover Accel: SMA3/SMA20 > 1.8 and |ret z|<1.

RSI Divergences: compare to n bars back (default 14).

How to use it (playbooks)

A) Daily scan workflow

Run on Daily for your VN watchlist.

Turn Composite (Σ) alert on with Σ≥2 or ≥3 to reduce noise.

When a bar fires Σ (or a fav combo like STL + BB↑), drop to 60-min to time entries.

B) Breakout quality check

Look for 52W together with BB↑, TAC, and OBV.

If WER/ND appear near highs → downgrade the breakout.

C) Spring/UTAD reversals

If SPR fires near major support and RBd confirms → long bias with stop below spring low.

If UTD + WER/RS↑ near resistance → short/fade with stop above UTAD high.

D) Accumulation basing

During bases, you want CLS, OBV, TGV, STL, NR7.

A pocket pivot (PIV) can be your early add; manage risk below base lows.

Tuning tips

Too many signals? Raise stealthVolZ to 5.5–6, evrVolZ to 3.5, use Σ≥3.

Fast movers? Lower bbwZthr to −1.0 (less strict squeeze), keep trend filter on.

Illiquid tickers? Keep MAD z-scores on, increase lookbacks (e.g., lenZ=120).

Limitations & good habits

First lenZ bars on a new symbol are less reliable (incomplete z-window).

Some ideas (VWAP magnet, close auction spikes, ETF/foreign flows, options skew) need intraday/external feeds — not included here.

Pine can’t “screen” across the whole market; set alerts or cycle your watchlist.

Quick troubleshooting

Compilation errors: make sure you’re on Pine v6; don’t nest functions in if blocks; each var int must be declared on its own line.

No shapes firing: check trend filter (maybe price is below SMA and you’re waiting for bullish signals), and verify thresholds aren’t too strict.