Loading live markets…
Model prob Consensus Market
Weather Intelligence
💡
Smart Take
Moderate signal: Models strongly agree this outcome. Market appears to be overpricing YES at 32% — consensus is 18%.
🌤️
Weather Edge Map
Hot (60+) Warm (30+) Low
Arbitrage Scanner
ProDetect cross-platform arbitrage when Kalshi + Polymarket prices create guaranteed profit windows.
Upgrade to Pro to unlockPaper Trading Portfolio
Sign in to track your P&L, win streaks, and performance.
City Forecasts
Explore weather contracts and 7-day forecast comparisons by city.
Embed Celsi on Your Site
Add a live weather edge widget to your blog or website.
<iframe src="https://celsi.vercel.app/embed" width="380" height="400" frameborder="0"></iframe>How to read Celsi
🎯
Edge Score (0-100)
How mispriced a contract is. Higher = bigger gap between what the market thinks and what weather models predict. 60+ is a strong signal.
📊
Consensus
The confidence-weighted average of all weather models (GFS, ECMWF, Ensemble). This is what Celsi thinks the true probability is.
💰
YES @ 52¢ / NO @ 48¢
The price to buy a contract. You pay the price in cents. If you're right, you get $1 back. YES @ 52¢ means the market says 52% chance.
📈
EV (Expected Value)
Your expected return if the models are right. +40% EV means you'd make 40¢ profit per $1 risked on average.
⭐
Buy YES / Buy NO
Celsi's recommendation. The starred button is the one the algorithm says has an edge. The explanation above tells you exactly why.
🔴🟡🟢
Model Dots
Each dot is a weather model's prediction. The diamond is the consensus. The dashed line is the market price. When dots are far from the line, there's an edge.
📏
Agreement Meter
Shows whether models agree. Green = high agreement (confident signal). Red = models disagree (riskier pick).
🔥
Signal Strength
Strong (score 60+), Moderate (40+), Mild (20+). Strong signals have high model agreement AND a big gap from the market.
Celsi pulls live contracts from Kalshi & Polymarket, compares them against GFS, ECMWF, and ensemble weather models, and highlights where the market is wrong. The algorithm learns from past predictions and improves over time.