Signal Engine

The Signal Engine is Nexxore's Market Intelligence Agent. It consumes real-time market data from the Market Data Agent and produces composite alpha trading signals using 4 independent detectors and a weighted scoring system.

---

Overview

The Signal Engine runs as a standalone Python/FastAPI microservice on port 3861. It pulls data from the Market Data Agent (port 3860), maintains rolling time windows, runs 4 signal detectors on each data update, and combines results into a 0–100 composite score.

Key Features

• 4 Alpha Signal Detectors — Momentum, volume, liquidity, and new pair detection

• Composite Scoring (0–100) — Weighted combination of all signals with classification

• Rolling Window Pipeline — Maintains 5m, 15m, 1h, and 24h time windows per pair

• Real-Time WebSocket — Live signal feed at /ws/signals/live

• REST API — Query top signals, strong trades, watchlist, per-token, per-chain

---

Signal Detectors

1. Momentum Spike (0–40 points)

Detects rapid price acceleration with volume confirmation.

Parameter	Threshold
Price change (5m)	> 5%
Volume (5m)	> 2× rolling average
Minimum liquidity	> $500,000

Strength breakdown:

• Price movement magnitude: 40%

• Volume ratio (current vs average): 35%

• Liquidity depth (pool size): 25%

---

2. Volume Breakout (0–30 points)

Detects when hourly volume significantly exceeds historical average.

Parameter	Threshold
Volume (1h)	> 3× average hourly volume from 24h
Minimum 24h volume	> $10,000
Minimum transactions	> 20 in 1h

Strength breakdown:

• Volume ratio (current vs historical): 45%

• Transaction velocity: 25%

• Buy pressure (buy ratio): 30%

---

3. Liquidity Inflow (0–20 points)

Detects significant capital entering a pool.

Parameter	Threshold
Liquidity increase	> 20% in 10-minute window
Minimum current liquidity	> $50,000

Strength breakdown:

• Inflow magnitude: 50%

• Resulting pool size: 30%

• Speed of inflow: 20%

---

4. New Pair Detection (0–10 points)

Identifies newly created pairs with meaningful early traction.

Parameter	Threshold
Pair age	< 24 hours
Minimum liquidity	> $100,000
Minimum volume (1h)	> $5,000

Strength breakdown:

• Liquidity depth: 35%

• Early volume: 30%

• Freshness (newer = higher): 20%

• Transaction activity: 15%

---

Scoring System

Each token receives a composite score from 0–100:

Total Score = momentum_score + volume_score + liquidity_score + new_pair_score
            = (0–40)        + (0–30)       + (0–20)          + (0–10)

Weights

Signal	Max Points	Weight
Momentum Spike	40	40%
Volume Breakout	30	30%
Liquidity Inflow	20	20%
New Pair	10	10%

Classification

Score Range	Classification	Action
> 70	🟢 Strong Trade	High-confidence alpha opportunity
50–70	🟡 Watchlist	Monitor for breakout confirmation
< 50	🔴 Ignore	No actionable signal

---

Architecture

Market Data Agent (:3860)
    │
    ▼
┌────────────────────────────┐
│  Data Pipeline             │
│  HTTP poll every 10s       │
│  Rolling Windows per pair  │
│  (5m / 15m / 1h / 24h)    │
└──────────┬─────────────────┘
           │
    ┌──────┴──────────────────────┐
    │      │       │              │
    ▼      ▼       ▼              ▼
┌──────┐ ┌──────┐ ┌──────────┐ ┌──────┐
│Momnm │ │Volume│ │Liquidity │ │New   │
│Spike │ │Break │ │Inflow    │ │Pair  │
│0-40  │ │0-30  │ │0-20      │ │0-10  │
└──┬───┘ └──┬───┘ └────┬─────┘ └──┬───┘
   │        │          │           │
   └────────┴──────────┴───────────┘
           │
           ▼
┌────────────────────────────┐
│  Signal Scorer             │
│  Weighted composite 0-100  │
│  Classification engine     │
└──────────┬─────────────────┘
           │
    ┌──────┴──────┐
    │             │
    ▼             ▼
┌─────────┐  ┌──────────┐
│ REST    │  │ WebSocket│
│ API     │  │ Live Feed│
│ :3861   │  │ /ws/...  │
└─────────┘  └──────────┘

---

API Endpoints

Method	Path	Description
GET	/api/v1/health	Health check + uptime
GET	/api/v1/stats	Service statistics
GET	/api/v1/signals/top	Top scored tokens (limit param)
GET	/api/v1/signals/strong-trades	Tokens with score > 70
GET	/api/v1/signals/watchlist	Tokens with score 50–70
GET	/api/v1/signals/{token_address}	Signal details for specific token
GET	/api/v1/signals/chain/{chain_id}	Signals filtered by chain
WS	/ws/signals/live	Live signal WebSocket feed

WebSocket Messages

The /ws/signals/live endpoint broadcasts three types of messages:

Type	Frequency	Content
signals_update	Every 5s	Top signals summary
signal_event	Real-time	Individual signal detections
heartbeat	Every 30s	Keep-alive ping

Client commands: send ping for pong, top for top signals, stats for service stats.

---

Quick Start

# Install dependencies
pip install -r services/signal-engine/requirements.txt

# Start Market Data Agent first (required)
python -m services.market-data.main &

# Start Signal Engine
python -m services.signal-engine.main

# Or with uvicorn
uvicorn services.signal-engine.main:app --port 3861 --reload

---

Data Flow

Dexscreener → Market Data Agent → Signal Engine → REST API + WebSocket
                                       │
                                       ├── On-Chain Analyst (research.html)
                                       ├── Execution Agents
                                       └── External Consumers

---

Configuration

Parameter	Default	Description
MARKET_DATA_URL	http://localhost:3860	Market Data Agent URL
SIGNAL_ENGINE_PORT	3861	Signal Engine port
SIGNAL_POLL_INTERVAL	10	Pipeline poll interval (seconds)
Momentum price threshold	5%	Price change 5m to trigger
Volume breakout multiplier	3×	Volume vs 24h average
Liquidity inflow threshold	20%	Liquidity increase in 10 min
New pair max age	24h	Maximum pair age
Strong trade threshold	>70	Score for strong trade
Watchlist threshold	50–70	Score for watchlist

---

Integration

The Signal Engine feeds into:

Consumer	Integration Method
On-Chain Analyst	REST API (Signals tab in research.html)
Execution Agents	REST + WebSocket
Strategy Builder	Signal consumption via API
Telegram Alerts	Webhook on strong_trade events (planned)

---

Next Steps

• Market Data Agent →

• On-Chain Analyst →

• Agent Overview →