PolyWeather is a specialized intelligence system built for Polymarket high-stakes participants. We aggregate top-tier meteorological sources, apply proprietary DEB (Dynamic Error Balancing) logic, and surface actionable shift signals at critical decision windows.

🏗️ Production Architecture

This project uses a decoupled production setup for reliability and iteration speed:

Frontend: A Next.js dashboard on Vercel with React component rendering.
Backend API: A FastAPI service on VPS for low-latency weather aggregation and analysis.
Bot & Alert Heartbeat: A Telegram Bot on VPS for minute-level scanning and push alerts.

🔗 Official Visit: polyweather-pro.vercel.app

🖼️ Preview & Interaction

📊 Deep Query View: DEB blended forecast + settlement probability + AI analysis context

🗺️ Omni-Dashboard: global station markers + nearby station context + right-side city intelligence panel

🚀 Core Features

📡 Full-Spectrum Collection
- Major Models: ECMWF, GFS, ICON, GEM, JMA, Open-Meteo, and city-level daily/hourly guidance.
- Observed Data: Aviation Weather / METAR as the primary observation source, plus Turkish MGM coverage for Ankara.
- City Specialization: 17130 (Ankara (Bölge/Center)) remains the Ankara lead station without replacing LTAC settlement observation.
⚖️ DEB Smart Blending
- Dynamic weighting based on city-level performance and current model spread.
📈 Market Data Integration
- Live Polymarket quotes, probabilities, and dynamic settlement bucket tracking.
- Automatic Market Edge and Spread calculation comparing DEB vs Market.
🧩 React Dashboard Runtime (v1.2)
- Fully internationalized (i18n) with English and Simplified Chinese support.
- Premium UI: Glassmorphism overlays, dynamic "Sonar Pulse" markers, and fluid loading states.
- Typed store + typed API client + Leaflet/Chart.js lifecycle wrappers.
- Today analysis workflow: open modal + freeze map motion.
🔔 Alert Engine
- Momentum Spike: Captures rapid short-window temperature slope changes.
- Forecast Breakthrough: Fires when observations break model envelopes plus margin.
- Advection Monitoring: Tracks warm/cold advection using lead-station behavior and wind direction.

🔐 Alert Logic Details

Trigger Name	Core Logic	Trading Value
Center Hit	Detects DEB trigger only at Ankara HQ `17130`	Highest priority signal, the "truth"
Momentum	30min temperature slope exceed threshold	Captures sudden weather fronts
Breakthrough	Pierces all model highs + margin	Captures high-volatility outlier events
Advection	Lead station rise + Wind match	Gain 20-40 minutes of lead time for execution

🧭 Current Data Logic

Primary observation source: Aviation Weather / METAR
Ankara enhancement:
- Settlement observation: LTAC / Esenboğa
- Official lead station: Ankara (Bölge/Center) / 17130
- Nearby station layer: Turkish MGM network (Ankara-specific preferred station ordering)
Other cities nearby layer:
- Production currently uses Aviation Weather METAR clusters
- U.S. cities may later receive Mesonet augmentation while METAR stays baseline
Frontend request optimization:
- Initial map temperatures preload via /api/city/{name}/summary
- City detail cache TTL = 5 minutes, revision probe avoids unnecessary refetch
- Map movements, panel toggles, and modal open/close do not trigger redundant requests
- Manual refresh always bypasses cache (force_refresh=true)

🏗️ System Architecture

graph TD
    subgraph "Client / Terminals"
        Web[Next.js React Web App]
        TG[Telegram Client]
    end

    subgraph "Edge Deployment (Vercel)"
        Web --> |BFF Routes| Fast[FastAPI API]
    end

    subgraph "Core Hub (VPS)"
        Fast --- |Shared Logic| Worker[Alert Engine / Worker]
        Bot[Telegram Bot] --- |Shared Logic| Worker
    end

    subgraph "External Sources"
        Worker --> |Pull| MGM[MGM Weather]
        Worker --> |Pull| METAR[Airport METAR]
        Worker --> |Pull| OM[Open-Meteo]
        Worker --> |Pull| MM[Multi-Model Integration]
    end

    Worker --> |Push Alert| TG
    Bot --> |Query| Worker

🛠️ Deployment

1. Backend / Bot (VPS)

# Pull source
git pull

# Environment
# Edit .env with TELEGRAM_BOT_TOKEN and other keys

# Launch
docker-compose up -d --build

2. Frontend (Vercel)

Set frontend as the Vercel root directory for automatic CI/CD.

💬 Bot Commands

Command	Description	Example
`/city`	Query real-time analysis for a city	`/city ankara`
`/deb`	View historical accuracy of DEB model	`/deb london`
`/top`	View activity leaderboard	`/top`
`/help`	Get detailed instructions	`/help`

Note

Commercialization: Current plans keep Web Dashboard ($5/mo) and Telegram Signal Channel ($1/mo) as the core entry offers. User entitlement and payment automation are tracked in docs/COMMERCIALIZATION.md.

Note

Frontend Model: Production rendering is now fully handled by React components under frontend/components/dashboard and hooks under frontend/hooks. Legacy static files are retained for reference, but no longer act as the main runtime path.

📅 Last Updated: 2026-03-10 🚀 Status: v1.2 Stable - React Dashboard with i18n & Polymarket Integration in Production

Tip

Production Note: The UI layout remains consistent while introducing full internationalization, market quote integration, and premium visual feedback (glassmorphism overlays, sonar markers).

Languages

Python 56.8%

TypeScript 37%

CSS 4.1%

JavaScript 1%

PLpgSQL 0.4%

Other 0.6%