MoneyRain v1.0

City 28 cities · Milan with +5 local models

Day ● = active Polymarket market

METAR—

Now obs—

Max today—

Updated—

Local time—

Forecast Max — Model aggregate

--°C

Median

Weighted

Range

Models

Expected Max — Calibrated obs + bias

--°C—

Observed floor

Projected peak

Peak hour

Obs source

Polymarket — Implied Max

--°C

Bucket top

Buckets

Market − models spread --

Confidence (model σ)

σ = -- --

<0.5 high · 0.5–1.0 good · 1.0–1.5 medium · >1.5 low

Top probabilities gauss · σ ensemble

Lowest temperature today · obs + forecast + market

Forecast

model aggregate

METAR low

—

Polymarket implied

—

Confidence (σ)

-- · --

Per-model breakdown

selected day · delta from mean · sorted by max desc ▼

Model	Center	Max	Δ mean	Weight	RMSE 6h	MAE 14d

Model accuracy (historical)

click to load · MAE/bias vs METAR · last 14 days ▼

Model	Center	N samples	MAE	Bias	MAE same-day	MAE 1-day	MAE 2-day
Click to load — reads from server cron history. Data starts populating ~24h after first deploy.

Spread scanner

today · raw model avg (no bias-corr) vs polymarket · |gap| ≥ 1°C / 2°F ▼

City	Forecast	Polymarket	Top bucket	Spread	Edge
Click to load — scans 28 cities (~5–10 sec)

Models via Open-Meteo · Polymarket via Gamma API (local proxy required from Italy)
For working Polymarket overlay, use serve.py included in the package

Per-model breakdown

Model accuracy (historical)

Spread scanner

Visual Crossing API key