AlphaAgent 🎯

Autonomous LLM agent for opportunity detection on prediction markets 🚧 Concept / design document — the code for this project is not published yet. This README describes the target design.

AlphaAgent is a multi-agent pipeline that continuously monitors prediction markets (Polymarket, Manifold) to detect high-value opportunities. It aggregates real-time news, historical market data and social sentiment to generate per-market confidence scores — fully automated, no manual analysis needed.

---

✨ Features

• Multi-source ingestion — news APIs, RSS feeds, Twitter/X, market history
• 3-agent architecture — Collector → Analyst → Decision-maker
• Confidence scoring — 0–100 score per market with reasoning
• Alert system — push notifications when score crosses threshold
• REST API — query scores programmatically from your bot
• SQLite persistence — full history of analyses and scores
• Async by design — concurrent market analysis, non-blocking

---

🧠 Agent Architecture

┌─────────────────────────────────────────────────────┐
│                  AlphaAgent Pipeline                │
│                                                     │
│  ┌─────────────┐                                    │
│  │  Collector  │ ← Polymarket API, news, RSS        │
│  │   Agent     │   Twitter/X, Reddit                │
│  └──────┬──────┘                                    │
│         ↓                                           │
│  ┌─────────────┐                                    │
│  │  Analyst    │ ← LLM: extract entities,           │
│  │   Agent     │   assess credibility, score signal │
│  └──────┬──────┘                                    │
│         ↓                                           │
│  ┌─────────────┐                                    │
│  │  Decision   │ ← LLM: final score + reasoning     │
│  │   Agent     │   + alert if threshold crossed     │
│  └─────────────┘                                    │
└─────────────────────────────────────────────────────┘

---

🚀 Quickstart

git clone https://github.com/high-altitude-ai/alpha-agent
cd alpha-agent

pip install -r requirements.txt

cp .env.example .env
# Add: GROQ_API_KEY, NEWS_API_KEY

python -m alphaagent.main --markets "US Election,Super Bowl"

---

📡 API

# Get score for a market
GET /score?slug=will-trump-win-2024

# Response
{
  "slug": "will-trump-win-2024",
  "score": 74,
  "confidence": "high",
  "reasoning": "Strong polling data corroborated by 3 independent sources...",
  "sources": ["reuters.com", "fivethirtyeight.com"],
  "updated_at": "2025-01-15T14:32:00Z"
}

# List top opportunities
GET /opportunities?min_score=70&limit=10

---

⚙️ Configuration

[agent]
model = "llama-3.3-70b-versatile"
provider = "groq"
scan_interval_seconds = 300
alert_threshold = 75

[sources]
news_api = true
rss_feeds = ["https://feeds.reuters.com/reuters/topNews"]
polymarket = true
manifold = true

[alerts]
webhook_url = ""     # Discord/Slack webhook

---

📁 Project Structure

alpha-agent/
├── alphaagent/
│   ├── agents/
│   │   ├── collector.py
│   │   ├── analyst.py
│   │   └── decision.py
│   ├── sources/
│   │   ├── polymarket.py
│   │   ├── news.py
│   │   └── social.py
│   ├── api.py
│   ├── db.py
│   └── main.py
├── tests/
├── .env.example
└── requirements.txt

---

📄 License

MIT — Guillaume Marty · high-altitude-ai