Skip to content

Latest commit

 

History

History
257 lines (211 loc) · 10.2 KB

File metadata and controls

257 lines (211 loc) · 10.2 KB

Agent–User Decision Workflow

From market updates to visualizations and reports — a step-by-step view of how Octopus agents and users collaborate for investment decisions (Aladdin-style decision support).


Overview

The platform combines continuous data ingestion, agent-driven analytics, and human decisions in a single loop: data flows in → agents analyze and recommend → the user decides → execution and reporting close the loop.

┌─────────────────────────────────────────────────────────────────────────────────┐
│  MARKET DATA  →  ENRICHMENT  →  SIGNALS & RISK  →  YOUR DECISION  →  EXECUTION   │
│       ↑              ↑                ↑                ↑                ↑       │
│     M1,M3,M9       M2,M5,M7         M4,M6           👤 You           M8,M11     │
│                                                          ↓                       │
│  REPORTS & VISUALIZATION  ←  LENS (M11)  ←  Results & attribution                │
└─────────────────────────────────────────────────────────────────────────────────┘

1. End-to-end pipeline (high level)

flowchart LR
    subgraph SOURCES["📡 Sources"]
        MKT[Markets]
        NEWS[News & Social]
        ALT[Alternative Data]
    end

    subgraph INGEST["🔄 Ingest & Store"]
        M1[Nexus M1]
        M2[Vault M2]
        M3[Pulse M3]
        M9[Echo M9]
    end

    subgraph ANALYZE["🧠 Analyze"]
        M5[Neuron M5]
        M7[Oracle M7]
        M4[Atlas M4]
        M6[Guardian M6]
    end

    subgraph DECIDE["👤 Decide"]
        USER[Trader]
    end

    subgraph EXECUTE["📤 Execute & Report"]
        M8[Shadow M8]
        M10[Chronicle M10]
        M11[Lens M11]
    end

    MKT & NEWS & ALT --> M1
    M1 --> M2
    M2 --> M3
    M1 --> M9
    M3 --> M5 & M7
    M9 --> M4
    M5 & M7 --> M4
    M4 --> M6
    M6 --> USER
    USER --> M8 & M10
    M8 --> M11
    M10 --> M11
    M11 --> REPORTS[Reports & Dashboards]
Loading

2. Decision workflow by phase (swimlane)

flowchart TB
    subgraph Phase1["Phase 1: Market updates"]
        A1[📡 Nexus M1: Ingest prices, news, alt data]
        A2[🗄️ Vault M2: Store & validate]
        A3[⚡ Pulse M3: Stream to platform]
        A4[💬 Echo M9: Sentiment scores]
        A1 --> A2 --> A3
        A1 --> A4
    end

    subgraph Phase2["Phase 2: Analytics & signals"]
        B1[🧠 Neuron M5: ML predictions]
        B2[🔮 Oracle M7: Price forecasts]
        B3[🎯 Atlas M4: Trading signals]
        B4[🛡️ Guardian M6: VaR, sizing, limits]
        B1 & B2 --> B3
        B3 --> B4
    end

    subgraph Phase3["Phase 3: Your decision"]
        C1[Dashboard & Command Center]
        C2[Review signals, risk, portfolio]
        C3[Approve / Reject / Modify]
        C1 --> C2 --> C3
    end

    subgraph Phase4["Phase 4: Execution & validation"]
        D1[📋 Shadow M8: Paper or live execution]
        D2[📜 Chronicle M10: Backtest if needed]
        D3[📊 Lens M11: Build views & reports]
        D1 --> D3
        D2 --> D3
    end

    Phase1 --> Phase2 --> Phase3 --> Phase4
    Phase4 -.->|Feedback| Phase1
Loading

3. Step-by-step: from market update to report

Step Phase Who / What Action Output
1 Market updates Nexus (M1) Ingest prices, fundamentals, news, social Normalized market & alternative data
2 Market updates Vault (M2) Store, validate, index Historical + real-time datasets
3 Market updates Pulse (M3) Stream live data, compute live metrics WebSocket feeds, live P&L
4 Market updates Echo (M9) Score news/social sentiment Sentiment signals per asset/theme
5 Analytics Neuron (M5) Run ML models (classification, prediction) Scores, predictions, regime labels
6 Analytics Oracle (M7) Time-series & price forecasting Price targets, scenarios
7 Signals Atlas (M4) Fuse signals, generate ideas Trading signals, strategy suggestions
8 Risk Guardian (M6) VaR, limits, position sizing, compliance Approved size, risk view, alerts
9 Decision 👤 You View Dashboard / Command Center; approve, reject, or adjust Your order / no-trade / strategy choice
10 Execution Shadow (M8) Paper or live execution, track fills Fills, position updates
11 Validation Chronicle (M10) Backtest strategy if needed Backtest report, metrics
12 Reporting Lens (M11) Build charts, dashboards, AI reports Visualizations & reports

4. User–agent interaction (sequence)

sequenceDiagram
    participant User as 👤 Trader
    participant UI as Dashboard / Command Center
    participant M4 as Atlas (Strategy)
    participant M6 as Guardian (Risk)
    participant M8 as Shadow (Execution)
    participant M11 as Lens (Reports)

    Note over UI,M11: Data & signals already in pipeline (M1–M7, M9)
    M4->>UI: Signals & strategy suggestions
    M6->>UI: Risk view, position sizing, limits
    UI->>User: Present options & risk
    User->>UI: Choose: approve / reject / modify
    UI->>M8: Send order (paper or live)
    M8->>UI: Fills & position update
    UI->>User: Confirmation
    User->>UI: Request report or dashboard
    UI->>M11: Build visualization / report
    M11->>UI: Charts, tables, insights
    UI->>User: Report & visualization
Loading

5. Where you see each agent in the UI

Agent Where in the platform What you get
Nexus (M1) Data Explorer, Command Center (data pipeline) Live feeds, export, data quality
Vault (M2) Data Explorer, exports Historical data, validated series
Pulse (M3) Dashboard, live charts, alerts Real-time prices, live analytics
Atlas (M4) Command Center → Strategies, Trading Bots Signals, strategy ideas, bot config
Neuron (M5) AI Models Predictions, model outputs
Guardian (M6) Command Center → Risk VaR, limits, risk dashboard
Oracle (M7) Options, predictions Price forecasts, scenarios
Shadow (M8) Paper trading, Portfolio (sim) Simulated execution, paper P&L
Echo (M9) Command Center (sentiment), Social Sentiment scores, news impact
Chronicle (M10) Command Center → Strategies → Backtesting Backtest results, strategy validation
Lens (M11) Reports, Visualization, Dashboard Charts, dashboards, AI report insights

6. Data flow: market data → visualization

flowchart TB
    subgraph Inputs["External inputs"]
        I1[Market data APIs]
        I2[News & social]
        I3[Alternative data]
    end

    subgraph Layer1["Layer 1: Ingest"]
        L1A[M1 Nexus]
        L1B[M2 Vault]
        L1C[M3 Pulse]
        L1D[M9 Echo]
    end

    subgraph Layer2["Layer 2: Intelligence"]
        L2A[M5 Neuron]
        L2B[M7 Oracle]
        L2C[M4 Atlas]
        L2D[M6 Guardian]
    end

    subgraph Layer3["Layer 3: Decision & execution"]
        L3A[User decision]
        L3B[M8 Shadow]
        L3C[M10 Chronicle]
    end

    subgraph Layer4["Layer 4: Output"]
        L4A[M11 Lens]
        L4B[Dashboards]
        L4C[Reports]
        L4D[Charts]
    end

    I1 & I2 & I3 --> L1A
    L1A --> L1B & L1C & L1D
    L1B & L1C & L1D --> L2A & L2B & L2C
    L2C --> L2D
    L2D --> L3A
    L3A --> L3B & L3C
    L3B & L3C --> L4A
    L4A --> L4B & L4C & L4D
Loading

7. One-page reference: “From market update to report”

  MARKET UPDATES          ENRICHMENT              SIGNALS & RISK           YOUR DECISION           EXECUTION & REPORT
  ───────────────        ───────────             ───────────────          ─────────────           ───────────────────

  Prices, news,          Stored & validated       Fused into signals       You see:                Orders executed
  alternative data  →    (Vault M2)          →   (Atlas M4)           →   risk (Guardian M6)  →   (Shadow M8)
       │                        │                        │                        │                        │
  Nexus M1, Pulse M3,     Neuron M5,             Atlas M4,              Command Center,          Chronicle M10
  Echo M9                  Oracle M7              Guardian M6             Dashboard                 backtests
       │                        │                        │                        │                        │
       └────────────────────────┴────────────────────────┴────────────────────────┴────────────────────────┘
                                                                                              │
                                                                                              ▼
                                                                                    Lens M11 → Visualizations
                                                                                              & Reports

Summary

  • Agents handle: ingestion (M1, M2, M3, M9), analytics (M5, M7), signals and risk (M4, M6), execution and backtest (M8, M10), and reporting (M11).
  • You decide: approve, reject, or modify in the Command Center and Dashboard, using agent outputs and risk views.
  • Flow: Market updates → enrichment → signals & risk → your decision → execution → visualizations and reports (Aladdin-style decision support).

For technical orchestration details, see AI-Agents and Architecture.