The most advanced browser-based 3D order book visualization platform ever built. Real-time depth surfaces, persistent trade bubbles, AI-powered pattern detection, multi-exchange execution, and quantitative microstructure analysis — all in a single HTML file.
3D NEXUS META transforms raw order book data into an immersive 3D battleground where every trade, every wall, and every manipulation pattern becomes visible.
3D NEXUS META runs as a single HTML file. No installation, no build pipeline, no dependencies to install.
Double-click the 3D_NEXUS_META.html file in your browser. Chrome or Edge recommended for best WebGL performance. The file works directly from file:// protocol — a CORS proxy chain handles REST API calls automatically.
The loading screen shows "INITIALIZING NEXUS META..." for about 2.5 seconds while the Three.js scene, particle systems, and WebSocket connections boot up. Once done, you'll see the 3D order book surface start building.
Use the Exchange dropdown in the Controls panel (top-left) to choose your data source. Select a trading pair from the Trading Pair dropdown. The symbol list loads dynamically from each exchange's API.
Left-click drag to orbit the camera. Scroll to zoom. Right-click drag to pan. Press R to reset the camera to the default position. Press F for fullscreen.
Every panel is draggable (grab the dotted handle), resizable (bottom-right corner), and has an opacity slider. You can minimize any panel with the — button. Use COLLAPSE ALL to hide everything except Controls.
The core visualization: a scrolling 3D terrain built from real-time Level 2 order book data.
InstancedMesh of thin vertical bars. Each bar represents the resting liquidity at a specific price level at a specific point in time. The X-axis is price (bids left, asks right), the Y-axis is size (taller = more liquidity), and the Z-axis is time (front = now, back = past).
round((sliceMidPrice - currentMidPrice) / tickSize). This means the price line stays straight while the depth canyon scrolls diagonally — giving you an instant visual of how the book structure shifted relative to price movement.
| SETTING | DEFAULT | DESCRIPTION |
|---|---|---|
| Price Levels | 50 | Number of price levels on each side of mid-price. 50 = 100 total columns. Higher values show more of the book but cost more GPU. Range: 20–100. |
| Time Depth | 100 | Number of historical slices kept. 100 at 500ms update rate = 50 seconds of history. Higher values create a longer canyon. Range: 50–200. |
| Height Scale | 1.0 | Multiplier for bar height. Increase to make large walls tower dramatically. Decrease for flatter terrain. Range: 0.1–10.0. |
| Update Rate | 500ms | How often the surface adds a new front row. 500ms is balanced. 100ms gives ultra-fast updates (⚡ shown). 2000ms gives slow-motion analysis. Range: 10–2000ms. |
| Tick Size | 0.01 | Price granularity for aggregation. Each surface column covers one tick. Smaller tick = finer resolution but potentially noisier. Larger tick = smoother aggregation. Auto-set for MT5 symbols from broker metadata. |
| Bubble Scale | 1.0 | Multiplier for trade bubble radius. Increase to make bubbles more prominent. Decrease for subtler trades. Range: 0.1–5.0. |
11 exchanges and data sources, each with native WebSocket protocol implementation.
ws://localhost:5555 for MT5, ws://localhost:5556 for Rithmic). The bridge polls DOM data, streams ticks, and routes orders to your MetaTrader 5 or Rithmic terminal. Symbols are loaded dynamically from your broker's offering — including case-sensitive names like ENQc1 on AvaTrade.
Every executed trade becomes a physical sphere falling from the sky onto the order book surface.
sqrt(relativeSize), where relativeSize = trade quantity / running average quantity. The Bubble Scale slider multiplies this radius.
Four microstructure patterns detected in real-time, visualized as 3D arrows on the surface and logged in the Scalp Signals panel.
A large wall near the mid-price holds steady while price pushes into it across multiple ticks. The wall doesn't shrink — it's absorbing the aggressive flow. BUY signal when bid walls absorb selling pressure. SELL signal when ask walls absorb buying pressure.
Detection: Checks for cumulative wall strength > 8 across the near-mid zone (20% of visible levels), combined with adverse price movement and delta shift.
A price level repeatedly fills, drains, and refills faster than normal. The institution is hiding size behind an iceberg algorithm that replenishes the visible level after each execution.
Detection: Pattern: big (t-3) → dropped <40% (t-2) → refilled (t-1). Price must be holding near the level (not breaking through), confirming it's a refill, not a pull.
Multiple walls on one side are melting (rapidly losing size) without being traded through. Supply or demand is exhausting — the opposite side is about to take control.
Detection: At least 2 levels with decay >1.5 from t-2, cumulative decay >5, and book imbalance confirming directional shift (>0.1).
A sudden burst of aggressive volume on one side (delta acceleration), confirmed by book imbalance. This is the classic "momentum ignition" pattern — a large player sweeping the book.
Detection: Delta short-term average minus long-term average >0.15, current delta >0.2, and same-direction book imbalance >0.05.
Three academic-grade microstructure metrics running in real-time on every trade.
>55% WARN · >65% DANGER · >80% TOXIC. At TOXIC, only delta_spike signals are allowed through.
NORMAL · INFORMED (R²>0.08) · TOXIC (R²>0.15 + λ>0.002). Reported in basis points.
<0.45 CALM · 0.45–0.70 NORMAL · 0.70–0.85 HOT · >0.85 CASCADE. CASCADE = herding behavior, most trades are chain reactions.
| OUTPUT | EFFECT |
|---|---|
| Confidence × | Multiplies signal confidence. Reduced by high VPIN and high Kyle R². Boosted by Hawkes CALM (quieter = more reliable signals). Range: 0.3–1.5. |
| Cooldown × | Multiplies signal cooldown ticks. Shortened during HOT/CASCADE markets (more signals OK). Lengthened during TOXIC flow (fewer, safer signals). Range: 0.3–3.0. |
| Spread Alert | Binary flag: "WIDEN" when VPIN > danger threshold or Kyle regime = TOXIC. Signals that you should be cautious about taking positions. |
| Composite Regime | Overall market regime: CALM → NORMAL → ELEVATED → DANGER → TOXIC. Displayed as a colored badge on the Quant Engine panel. |
Execute trades directly from the 3D visualization with TP/SL management.
modify_position — the 2-step approach ensures the actual fill price is used for calculation. For LIMIT/STOP, TP/SL are computed from the order price.
CLOSE ALL button sends a close-all command to MT5/Rithmic. POSITIONS button queries current open positions.
Automated execution triggered by scalp signals — the AI trades for you.
| SETTING | DEFAULT | DESCRIPTION |
|---|---|---|
| MIN CONF % | 65 | Minimum signal confidence required to trigger an auto-trade. Signals below this threshold are ignored. Re-read every signal. |
| MAX POS | 1 | Maximum simultaneous open positions. Once hit, new signals are ignored until a position closes. |
| TP (TICKS) | 20 | Take-profit distance in native ticks. Applied via modify_position after fill. |
| SL (TICKS) | 10 | Stop-loss distance in native ticks. Applied via modify_position after fill. |
| Signal Types | ALL ON | Checkboxes for each signal type: ABSORB, ICEBERG, EXHAUST, DELTA. Disable types you don't trust. |
| Cooldown | 15s | Minimum time between auto-trades. Prevents rapid-fire entries during volatile moments. |
ACTIVE → settings are read → signal fires → MARKET order sent to MT5 → fill confirmed → TP/SL applied via modify_position → position monitored → closed by MT5 natively when TP/SL hits → stats updated.
Complete reference for every control available in the interface.
| CONTROL | LOCATION | DESCRIPTION |
|---|---|---|
| Exchange | Controls | Dropdown selecting the data source. Changing this reloads the symbol list and reconnects all WebSockets. |
| Symbol Search | Controls | Type-to-filter the symbol dropdown. Case-insensitive. Works across all exchanges. |
| Trading Pair | Controls | Symbol dropdown. Selection triggers WebSocket reconnection and book reset. |
| Show Trade SIZE in Bubbles | Controls | Checkbox. Shows a text sprite inside each bubble with the trade quantity. |
| Show Depth SIZE on Bars | Controls | Checkbox. Shows persistent size labels above each depth bar, drifting backward with the surface. |
| FOOTPRINT | Toggle Row | Toggle the Footprint Volume Bar (bid×ask horizontal bars above the surface). Key: P |
| HEATMAP | Toggle Row | Toggle the Bookmap-style heatmap overlay. Key: H |
| CARPET/WALL | Toggle Row | Switch heatmap between floor carpet mode and back-wall mode. Key: M |
| LIQUID | Toggle Row | Toggle the Liquid Metal surface deformation (wave physics). Key: L |
| FP BAR | Toggle Row | Toggle the Footprint Volume Bar. Key: B |
| QUANT | Toggle Row | Toggle the Quant Engine panel (VPIN/Lambda/Hawkes). Key: Q |
| VOL PRO | Toggle Row | Toggle the 3D Volume Profile with VPOC laser. Key: G |
| VELOCITY | Toggle Row | Toggle the Trade Velocity Heatmap band. Key: T |
| TEMPLATE | Controls | Opens the visual theme selector: NEXUS CYBER, TERMINAL PRO, or AURORA DEEP. |
| COLLAPSE ALL | Controls | Minimize all panels to their title bars. Click again to expand all. |
| RESET (↺) | Info Panel | Reset all panel positions to default layout. |
| CLEAN MEM | Info Panel | Flush all 3D memory: bubbles, signals, particles, holograms, price labels. Rebuilds the surface. Use if performance drops. |
Three distinct visual themes, each transforming the entire interface and 3D scene.
Quick reference for all keyboard controls.
The cyan price line connects each trade bubble chronologically, creating a pure trade-by-trade price trail across the 3D surface. Unlike traditional mid-price lines, this shows actual execution prices — no averaging. A glowing sphere pulses at the latest trade. A purple shadow line runs slightly below for depth perception.
The line sits above the surface, with Y-position dynamically adjusted based on the surface height at each trade's grid position. This means the price trail naturally flows over the order book terrain, rising over whale walls and dipping into valleys.
A horizontal bar displayed above the order book surface showing executed Bid × Ask volume at each price level for the current time slice. Left bars (red) = sell volume hitting the bid. Right bars (green) = buy volume lifting the ask. The dominant side is highlighted brighter.
Numeric labels appear above the most active levels showing exact volume with an imbalance indicator (colored bar showing buy/sell split and net delta). The footprint data accumulates from the aggTrade stream and resets with each new surface slice.
Toggle with B key or the FP BAR button in the toggle row.
A Bookmap-style 2D heatmap rendered as a texture on a plane in the 3D scene. Each pixel row represents one time slice, each column represents a price level. Color intensity maps to resting liquidity size using a custom LUT: deep navy (empty) → teal (light) → yellow (medium) → orange (heavy) → red/white (extreme).
Two display modes: CARPET (flat on the ground below the surface, like a reflection) and WALL (vertical backdrop behind the surface). A cyan pixel line marks the mid-price in each row.
Toggle with H, switch mode with M.
A 4-layer physics simulation that makes the order book surface behave like a viscous liquid. Large trades create ripple waves that propagate across the surface. Book changes cause the bars to overshoot and settle (spring-damper physics). Neighbor cells pull each other (surface tension), and the entire surface breathes with a subtle sinusoidal pulsation.
The effect is subtle by design (V5.1 tuning) — it adds organic life to the surface without overwhelming the data. Toggle with L.
A heuristic pattern recognition engine that detects 5 smart-money microstructure signals: Iceberg Orders (hidden refilling liquidity), Spoofing (fake walls placed and pulled), Absorption (walls eating flow without shrinking), Exhaustion (walls melting rapidly), and Smart Money Accumulation/Distribution (clusters of large walls holding while price drifts).
Each detection spawns a 3D hologram on the surface (arrow + glowing ring + text label), adds an entry to the Oracle feed panel, and optionally triggers voice commentary via Web Speech API. The voice uses a deep, cinematic tone (pitch 0.7, rate 0.88). Toggle voice with the 🔊/🔇 button.
The Oracle uses rolling window buffers of the last 20 surface slices (levelHistory) to detect temporal patterns — it needs at least 3 ticks of data before firing.
Five autonomous AI agents powered by Groq LLMs orbit the scene as 3D drones (octahedron + torus halo). Each agent receives a real-time market snapshot and returns a structured JSON verdict.
Agents:
⚡ DELTA HUNTER — Cumulative delta momentum analysis. Predicts BULL/BEAR/NEUTRAL.
🧊 ICEBERG ASSASSIN — Hidden iceberg order detection. Reports DETECTED/CLEAR/SUSPECTED.
👻 SPOOF WARDEN — Spoofing and layering detection. Reports SPOOFING/CLEAN/SUSPICIOUS.
🛡️ ABSORPTION SAGE — Absorption vs exhaustion classification. Reports ACCUMULATION/DISTRIBUTION/EXHAUSTION/NEUTRAL.
👁️ CONSENSUS OVERLORD — Synthesizes all 4 agents into a final directional verdict with probability.
To activate: paste your Groq API key (starts with gsk_) and click ACTIVATE. Models are resolved dynamically from /v1/models — the system picks the best available fast model (8B) for agents and the best large model (70B) for the Overlord.
A real-time tape showing every executed trade with columns: TIME, PRICE, SIZE, SIDE (BUY/SELL), IMPACT (relative to running average: sm/MED/LARGE/🐋 WHALE), and VOL PROFILE (stacked buy/sell volume at each price level as a mini bar chart).
Rows are color-coded with a heatmap: small trades are dim, larger trades get brighter. The session buy/sell volume ratio is shown at the top. Pause the tape with the ⏸ button to analyze prints without scrolling.
Keeps up to 300 prints in memory, renders up to 80 DOM rows. Volume profile bars update every 2 seconds across all visible rows.
A session-accumulated volume histogram displayed as horizontal 3D bars on the right side of the surface. Each bar represents total executed volume at a price level, color-coded by buy/sell dominance: green (buyer-dominated) → purple (balanced) → red (seller-dominated). The VPOC (Volume Point of Control — the price with the highest volume) is highlighted in gold with a horizontal laser line extending across the entire scene.
A session stats label shows total buy/sell volume, percentage split, imbalance bar, and level count. Toggle with G.
A color band at the front of the surface showing the speed of tape (trades per second) over a rolling 3-second window. The band uses a cinema-grade color ramp: deep navy (dead calm) → cool blue → cyan/teal → yellow/orange → red/white (burst). An adaptive max tracks the peak rate.
When velocity exceeds 80% of the adaptive max (minimum 8 trades/sec), the Oracle fires a "TAPE BOMB" alert — likely an institutional sweep or news event. Toggle with T.
A sliding notification feed on the right side showing three types of events:
🐋 WHALE WALL — A single price level with size > 4.0 (normalized). Gold flash, whale alert sound, explosion particles.
AGGRESSOR — A level with size > 2.5. Screen flash, kill hit sound.
ABSORPTION — Over 60% of one side's levels have size > 0.8. Explosion at center.
Events are throttled by a cooldown system to prevent spam. Each entry shows type, side (buy/sell border color), size, price, and timestamp. Entries auto-fade after 5 seconds.
Every panel in the interface supports: Drag (grab the dotted handle at the top), Resize (drag the bottom-right corner), Opacity (slider with ◑ icon), Minimize (click — button in the handle), and Position persistence (saved to localStorage, survives browser restart).
Panels: 3D NEXUS (info), CONTROLS, ORACLE, SWARM, EXECUTE, API KEYS, TIME & SALES, SCALP SIGNALS, DELTA (power gauge), QUANT ENGINE. All draggable, all resizable, all minimizable.
Use ↺ RESET in the Info panel to reset all positions to the responsive default layout. The layout is computed based on your viewport dimensions.
For MetaTrader 5 (CME Futures, Forex, CFD), a Python bridge server connects NEXUS META to your MT5 terminal via WebSocket on port 5555. The bridge handles: DOM polling (order book snapshots), tick streaming (trade events), symbol discovery (from your broker's offering), and order execution (MARKET, LIMIT, STOP with TP/SL).
Setup: Run python mt5_bridge.py → Select "MT5 Local (Futures / CFD)" in the Exchange dropdown → Symbols load automatically → Select your instrument → Data flows.
MT5 symbols are case-sensitive (e.g., ENQc1 on AvaTrade). The bridge preserves original case for all subscribe, order, and display operations. Tick size is auto-detected from broker metadata but can be manually overridden without disconnecting.
For direct CME/CBOT/NYMEX market data and execution via Rithmic. Same bridge architecture as MT5 but on port 5556. Supports ES, NQ, YM, RTY (and Micros), CL, GC, SI, ZB, ZN, currency futures, and agricultural futures.
Select "Rithmic Futures (CME/CBOT/NYMEX)" in the Exchange dropdown. A fallback static symbol list provides common CME instruments if the bridge isn't running yet.
For Binance/Bitget execution, paste your API key and secret in the API Credentials panel. Keys are stored in sessionStorage only — they persist within the browser tab but are cleared when you close it. Keys are never sent to any server except the exchange itself.