PolyWeatherLab
What Polymarket priced, what the weather models said, and what actually happened.

Seattle highs

The market this page measures: highest temperature in Seattle on 28 September 2026 on Polymarket
Whatever the market priced for 2026-09-28, what every weather model forecasts for the same day, and what actually happened, all in °F. The gap column is how far a model sits from the market: above the market it ran warm, below it ran cool. Resolves on station KSEA.
|||
models say
65.6 °F
avg of 13 models
market's pick
66-67 °F
54.00% on this bucket
gap
+0.42 °F
spread 6.5 °F
actual
not resolved yet
A temperature difference scales rather than shifts: a gap of 1 °C reads as 1.8 °F. The colour bands still use the Celsius thresholds, so a row can look large here and still count as a small disagreement.

Market odds for 2026-09-28

What Polymarket is paying for each temperature. Each price is the market's probability that the day lands in that bucket, so they add up to about 100%. Latest reading 2026-09-28 09:21 UTC, shown in °F.
swipe the table sideways for every column
temperature °Fodds
58-59 °F0.15%
60-61 °F1.85%
62-63 °F3.35%
64-65 °F28.50%
66-67 °F54.00%
68-69 °F11.00%
70-71 °F1.55%
72-73 °F0.25%
74-75 °F0.05%

Every forecast for 2026-09-28

Each model's newest reading for this day, in °F, taken 2026-09-28 06:40 to 09:00 UTC. Gap is the difference from the market, in degrees: a model above the market ran warm, one below ran cool.
swipe the table sideways for every column
modelforecast °Fgap vs market °F
THE MARKET66.0––
best match62.2-3.8
ncep hrrr conus62.2-3.8
cmc gem rdps63.1-2.9
dwd icon global63.7-2.3
ecmwf ifs02563.9-2.1
jma gsm64.2-1.8
ncep nbm conus66.6+0.6
meteofrance arpege world66.7+0.7
cmc gem hrdps67.1+1.1
ukmo seamless67.6+1.6
cma grapes global68.2+2.2
ncep gfs global68.2+2.2
ncep nam conus68.7+2.7

How the forecasts moved

Every reading we hold for 2026-09-28, newest column first, in °F. The bracket under each time says how long before the day that reading was taken (or after, once the day has started), so the table reads right to left as the day approached. A model still moving in the last day is a model that has not settled, and that is the one to distrust. Older readings are kept but not shown.
swipe the table sideways for every column
model09-28 09:00 (2h after)09-28 08:00 (1h after)09-28 06:40 (0h before)09-28 06:07 (1h before)09-28 06:01 (1h before)09-28 05:30 (1h before)
THE MARKET66.066.166.366.366.366.3
best match62.2–63.362.462.4–
cma grapes global––68.268.2––
cmc gem hrdps––67.167.1––
cmc gem rdps––63.163.1––
dwd icon global––63.763.7––
ecmwf ifs025–63.964.664.6––
jma gsm––64.264.2––
meteofrance arpege world––66.766.7––
ncep gfs global––68.268.2–68.2
ncep hrrr conus62.2–63.362.462.4–
ncep nam conus68.7–66.466.4––
ncep nbm conus66.6–64.964.9––
ukmo seamless––67.667.6––