DCScout — AI-Powered Trading Intelligence Platform

Browser Use Web Agents Hackathon 2026

DCScout deploys 50+ parallel browser agents to generate institutional-grade trading intelligence across three asset classes — data center equities, credit markets, and European power — replacing analyst teams with autonomous web agents that deliver actionable signals in minutes.

Live Dashboard | Zero API keys required for power data

---

What It Does

1. Data Center Intelligence (9 agents, ~5 min)

Replaces a $5,000, 2-week analyst site visit with a 5-minute automated intelligence brief.

Agent	What it does	Source
Financial	Stock price, analyst ratings, short interest	Finviz
Industry News	Expansion announcements, capacity additions	DatacenterDynamics, Reuters
Job Postings	Open roles near data center sites	LinkedIn
Building Permits	County permit filings	Government records
Power Deals	Utility agreements, MW capacity	News + filings
Local News	Community impact, local developments	Google News
Social Sentiment	Bullish/bearish analysis	X/Twitter
Real Estate	Land sales, price per acre	Zillow, Redfin
Satellite Recon	Construction activity from imagery	Google Maps

2. European Power Trading (24 agents + 12 live APIs)

Real-time intelligence for battery energy storage (BESS) operators and power traders.

Browser Agents scan EPEX SPOT, ENTSO-E, Windy.com, EEX, and 20+ sources.

12 Free Live APIs (no keys needed):

Source	Data	Provider
EPEX Day-Ahead Prices	Hourly EUR/MWh across 6 bidding zones	Energy-Charts (Fraunhofer ISE)
Generation Mix	Wind, solar, coal, gas, nuclear by hour	Energy-Charts
Installed Capacity	GW by source type (2002-2026)	Energy-Charts
Renewable Forecast	Predicted renewable % next 24h+	Energy-Charts
Cross-Border Flows	Physical flows with 11 neighboring countries	Energy-Charts
Grid Signal	Traffic light (green/yellow/red)	Energy-Charts
Wind Speed	10m + 100m hub height forecasts	Open-Meteo
Solar Radiation	Shortwave radiation + cloud cover	Open-Meteo
Grid Generation	Quarter-hourly by source	SMARD (Bundesnetzagentur)
Grid Load	Total + residual load	SMARD
FCR Prices	Frequency containment reserve blocks	regelleistung.net
aFRR Prices	Automatic frequency restoration reserve	regelleistung.net

Trading signals: SHORT solar floods, LONG evening ramps, FCR block arbitrage.

3. Credit Trading (6 agents)

CDS-bond basis arbitrage detection with FINRA TRACE, DTCC, and Fed curves.

4. Bloomberg-Grade Research (11 agents)

SEC filings, 13F ownership, options flow, earnings transcripts, supply chain mapping, dark pool volume, Google Trends, Glassdoor sentiment.

---

Quick Start

export BROWSER_USE_API_KEY=bu_...
pip install -r requirements.txt

# Data center intelligence (9 agents)
python3 main.py --target equinix

# European power trading (24 agents + 12 APIs)
python3 main.py --mode power

# Power trading 24/7 daemon (scans every 15 min)
python3 main.py --mode power --daemon

# Credit basis trading
python3 main.py --mode basis --target "Goldman Sachs"

# Interactive dashboard
streamlit run app.py

# Quick test (2 agents, ~30 seconds)
python3 main.py --quick

---

Sample Output

Data Center Brief

============================================================
  DCScout INTELLIGENCE BRIEF
============================================================
  Target:     Equinix (EQIX)
  Market:     Northern Virginia
  Signal:     LONG
  Confidence: 100%
  Thesis:     Equinix shows strong expansion signals across 8/8 data sources.
============================================================
  Agents:     9  |  Time: 307s  |  Cost: $0.00
  Analyst equivalent: ~$5,000 (2-week site visit)
============================================================

Power Trading Signal

============================================================
  EUROPEAN POWER INTELLIGENCE
============================================================
  Country:    Germany (DE)
  Direction:  SHORT next 4h
  Conviction: 87%
  Wind:       26.6 GW (peak) | Solar: 14.2 GW
  EPEX DA:    €38.50/MWh avg | Spread: €92.10
  FCR:        €45.61/MW/block | Annual: €99,893/MW
============================================================
  12 live sources | 5 providers | 0 API keys
============================================================

---

Real Intelligence Gathered

Data Centers (Feb 28, 2026)

• $974.26 Equinix stock, Strong Buy consensus
• £3.9B UK campus acquisition (250MW+)
• $700M Amazon land deal (189 acres, Bristow)
• $615M SDC Capital Partners ($6.3M/acre — Loudoun County record)
• 40-47 GW data center pipeline in Northern Virginia

European Power (Live)

• Cross-referenced accuracy: Energy-Charts vs SMARD within 0.1% (26,641 vs 26,616 MW wind)
• FCR revenue validated against FlexPower benchmark (€99.9K vs €134K full-stack)
• 6 bidding zones: DE, AT, FR, NL, BE, CH

---

Architecture

main.py              → CLI entry point (7 modes)
orchestrator.py      → Parallel agent dispatch (asyncio.gather + threading)
agents.py            → 50+ browser agent definitions (Browser Use Cloud SDK)
models.py            → Pydantic schemas (50+ data types)
app.py               → Streamlit interactive dashboard

# Data Center Intelligence
dashboard.py         → HTML report generator

# European Power Trading
energy_charts_api.py → 6 Energy-Charts endpoints (Fraunhofer ISE)
smard_api.py         → SMARD Bundesnetzagentur grid data
balancing_api.py     → FCR/aFRR from regelleistung.net
weather_api.py       → Open-Meteo wind + solar forecasts
power_map_dashboard.py → Map dashboard (Leaflet + Chart.js)
power_dashboard.py   → EPEX price dashboard

# Credit Trading
credit_scorer.py     → 6-factor credit scoring
credit_universe.py   → 35+ credit definitions

# Infrastructure
convex_sync.py       → Real-time Convex backend sync
convex/              → Serverless database (scans, prices, signals)
deploy/              → Cloudflare Pages deployment

---

Key Technical Decisions

• True parallelism: asyncio.to_thread() wraps synchronous Browser Use SDK for real concurrent execution across 50+ agents
• Zero-key data pipeline: 12 live European market data sources requiring no API keys or registration
• Cross-reference verification: Multiple independent government sources validated against each other (Energy-Charts vs SMARD = 0.1% variance)
• Self-contained HTML: Dashboards are single files with embedded CSS/JS, deployable anywhere
• Convex real-time backend: HTTP ingestion with serverless mutations for live data persistence
• FCR block pricing: Settlement prices correctly handled as EUR/MW per 4-hour block (not hourly)

---

Data Accuracy

All data is sourced from official government and institutional providers:

Provider	Authority	Update Frequency
Energy-Charts	Fraunhofer ISE (German research institute)	15-minute
SMARD	Bundesnetzagentur (German federal regulator)	Quarter-hourly
regelleistung.net	Joint TSO platform (50Hertz, Amprion, TenneT, TransnetBW)	Daily tenders
Open-Meteo	ECMWF/DWD weather models	Hourly
EPEX SPOT	European power exchange (via Energy-Charts)	Hourly/15-min

---

Built With

• Browser Use Cloud — Browser agent infrastructure
• Python 3.11+ / asyncio / Pydantic
• Streamlit — Interactive dashboard
• Convex — Real-time serverless backend
• Cloudflare Pages — Static dashboard hosting
• Chart.js / Leaflet.js — Data visualization