Forecast accuracy · measured in the open

Know which forecast to trust.

Every major weather model, scored against what actually happened — city by city. No model wins everywhere. We tell you which one wins where, and we publish the misses too.

Scored against: NOAA · ECMWF · ICON · GFS | 19 models · every market, global · exact settlement station
The proof

The best model is a different model in every city.

We publish the exact settlement station and which of 19 models wins each city — graded in the open, misses shown. The public track record is being built; per-city hit rates appear once it exists.

City scoreboard — 1 day ahead

record start date: to be announced
Sample data

Record start date: to be announced

City-by-city hit rates are not shown until the new, hash-sealed record has run. Any figure shown before then is clearly labeled Sample data and does not count toward a claim.

What you get

Built for prediction-market traders.

The same measured forecasts behind every number — turned into the three things you actually trade on, across every Polymarket & Kalshi temperature market.

Best model

Which model to trust

We score the exact settlement station and tell you which of 19 models wins each city — misses shown.

Bin odds

Odds for every bin

A probability for each whole-degree temperature bin, so you price every outcome — not just the favourite.

The edge

Where the price is wrong

Our odds minus the live Polymarket / Kalshi price, ranked by expected value — the mispriced markets surface first.

Why us

Everyone shows their wins. We show the misses too.

Other tools quote one headline accuracy and stop. We publish every city, every model, graded in the open — including the ones we're bad at, like Denver. A desk sees exactly where the signal is strong before paying a cent.

That's the whole pitch: a track record you can check, not a claim you have to trust.

Early access

See the scoreboard before anyone else.

Drop your email. Get the free scoreboard first, and edge alerts the moment Pro launches.