PropBetEdgePREDICTIONS
Boston — photograph (Photo: Robert Linsdell from St. Andrews, Canada · CC BY 2.0 · via Wikimedia Commons)
WEATHER · FORECAST CHANGE

Where will it rain on Oct 4, 2026: PBE moves “Boston” up to 52% from 32%

New forecast guidance reached the model between 20:00 UTC and 23:00 UTC on Oct 3. Same model version, later data cutoff: here is exactly which inputs changed.

Published Model data as of Oct 3 23:00ZBy the PropBetEdge Predictions Desk1 min read

Photo: Robert Linsdell from St. Andrews, Canada · CC BY 2.0 · Wikimedia Commons

32%52%

Boston · pbe-weather-precip@1.1.0 · 20:00 UTC → 23:00 UTC

Move+20 pts
Market at capture28%
Inputs changed4primary
LIVE UPDATE · not part of the original story

Where will it rain on Oct 4, 2026?

Boston · RESOLVED

PBE · live intraday6%−46 pts since publication
KalshiNo two-sided quote
GapNo comparable quote

At KBOS: high so far 63.0°F · latest 57.2°F () · rain so far 0.00 in · STALE

RESEARCH PBE Oct 5 04:08Z · market

  1. UPDATE · Oct 4 18:18Z PBE 73% pbe-weather-precip-intraday@2.0.0 · data cutoff Oct 4 18:18Z
  2. UPDATE · Oct 4 21:03Z PBE 58% pbe-weather-precip-intraday@2.0.0 · data cutoff Oct 4 20:13Z
  3. UPDATE · Oct 4 23:03Z PBE 35% pbe-weather-precip-intraday@2.0.0 · data cutoff Oct 4 23:00Z
  4. UPDATE · Oct 5 01:03Z PBE 25% pbe-weather-precip-intraday@2.0.0 · data cutoff Oct 5 00:23Z
  5. UPDATE · Oct 5 03:03Z PBE 13% pbe-weather-precip-intraday@2.0.0 · data cutoff Oct 5 02:23Z
Open the live forecast →

Quick read

  • pbe-weather-precip@1.1.0 moved “Boston” from 32% to 52% (+20 pts).
  • Later data cutoff: Oct 3 17:00Z → Oct 3 23:00Z; same model version, so the move reflects inputs, not a model change.
  • Primary inputs changed: GFS MOS max 6-h chance of rain, National Blend max 6-h chance of rain, GFS MOS chance of rain in the window, National Blend chance of rain in the window.
  • The Kalshi mid captured with each snapshot went from 36% to 28%.

What changed in the inputs

Between the snapshot captured at Oct 3 20:00Z (data cutoff Oct 3 17:00Z) and the one at Oct 3 23:00Z (cutoff Oct 3 23:00Z), the same model version — pbe-weather-precip@1.1.0 — moved “Boston” from 32% to 52%. The model inputs that changed: GFS MOS max 6-h chance of rain from 16% to 34%; National Blend max 6-h chance of rain from 35% to 38%; GFS MOS chance of rain in the window from 28% to 52%; National Blend chance of rain in the window from 47% to 58%.

EVIDENCEStored model inputs, before and afterFrom the immutable feature snapshots attached to each forecast
InputBeforeAfterPrimary
GFS MOS max 6-h chance of rain16%34%yes
National Blend max 6-h chance of rain35%38%yes
GFS MOS chance of rain in the window28%52%yes
National Blend chance of rain in the window47%58%yes
Guidance lead time17h11h—
Units: as labeledAs of Oct 3 23:00ZSource: PBE feature snapshots

This page reports which stored inputs differ between two immutable forecasts. It does not claim why the underlying weather guidance changed.

The forecast path

PROBABILITY HISTORY“Boston”: pbe-weather-precip@1.1.0 snapshots vs the marketSame model version only; the dashed line is the stored Kalshi mid
0%25%50%75%100%Oct 3 18:31ZOct 3 20:45ZOct 3 23:00Z32%52%PBE pbe-weather-precip@1.1.0 snapshots 32% · Oct 3 18:31ZPBE pbe-weather-precip@1.1.0 snapshots 32% · Oct 3 20:00ZPBE pbe-weather-precip@1.1.0 snapshots 52% · Oct 3 23:00Z
Kalshi mid (stored observations)PBE pbe-weather-precip@1.1.0 snapshots
Units: % probabilityAs of Oct 3 23:00ZSource: PBE forecast archive; Kalshi
Data table
Time (UTC)SourceValue %
Oct 3 18:31ZPBE pbe-weather-precip@1.1.032
Oct 3 18:31ZKalshi mid32
Oct 3 19:00ZKalshi mid31
Oct 3 19:15ZKalshi mid36
Oct 3 20:00ZPBE pbe-weather-precip@1.1.032
Oct 3 20:00ZKalshi mid36
Oct 3 20:15ZKalshi mid28
Oct 3 20:30ZKalshi mid28
Oct 3 20:45ZKalshi mid28
Oct 3 21:00ZKalshi mid31
Oct 3 21:15ZKalshi mid33
Oct 3 21:30ZKalshi mid34
Oct 3 21:45ZKalshi mid34
Oct 3 22:00ZKalshi mid34
Oct 3 22:15ZKalshi mid32
Oct 3 22:30ZKalshi mid32
Oct 3 22:45ZKalshi mid29
Oct 3 23:00ZPBE pbe-weather-precip@1.1.052
Oct 3 23:00ZKalshi mid28

The Kalshi mid captured with each snapshot went from 36% to 28%.

The two snapshots

EarlierLater
CapturedOct 3 20:00ZOct 3 23:00Z
Data cutoffOct 3 17:00ZOct 3 23:00Z
Guidance runsNWS GFS MOS 2026-10-03T12Z, NWS National Blend of Models 2026-10-03T12ZNWS GFS MOS 2026-10-03T18Z, NWS National Blend of Models 2026-10-03T18Z
PBE32%52%
Market at capture36%28%

Evidence & method

What this story is based on. Every value above was read from these immutable records.

Boston
Model
pbe-weather-precip@1.1.0 · RESEARCH
Snapshot
8aaefaa1-6b97-4ff4-a953-5f84c002d200
Compared with
f85d30bb-6191-40ed-883a-2877b3f6c7b0
Captured
Oct 3 23:00Z
Data cutoff
Oct 3 23:00Z
Market observation
28% · captured with the snapshot
Source families
NOAA/NWS via Iowa Environmental Mesonet archive · NOAA Regional Climate Centers ACIS · NOAA/NWS api.weather.gov
RESOLUTION RULE
Resolves on
The Weather Company (weather.com/kalshi) — The Weather Company daily precipitation for CLIBOS (weather.com/kalshi)
Independent check
NWS Daily Climate Report CLIBOS (WFO BOX); NOAA RCC-ACIS USW00014739
Measurement
Total liquid-equivalent precipitation for the climate day 2026-10-04, 00:00-24:00 local standard time (UTC-5)
Rounding
Hundredths of an inch; trace (T) counts as 0.00
Exceptions
Missing daily precipitation value counts as 0 inches (resolves NO) · Preliminary data may be revised; the exchange may hold expiration until a non-erroneous revision, else last fair price · Venue event strike_date (2026-10-05T03:59:59Z) is not the climate-day boundary; the reported daily value governs
If the total precipitation at CLIBOS in Boston in Oct 4, 2026 is strictly greater than 0 inches, then the market resolves to Yes.
MODEL LIMITATIONS
  • Forecasts are published only before the climate day opens (no intraday observations in v1)
  • IEM NBM archive gaps -> GFS-only fallback tier
  • Settles on The Weather Company values; verified against NWS CLI 112/112

Weather · measurable rain — inputs: NWS GFS MOS + National Blend of Models 6-h PoP for the exact climate day, 1991-2020 station climatology (ACIS). This model on the research board →

Market prices are a benchmark only and never enter a PropBetEdge model. Research-stage probabilities; not advice.