Forecast NO — High MAE Cities

NWS forecast bracket misses ~75-89% of the time in high miss-rate cities. Each city has its own NO entry cap (= miss rate − 10pp safety margin).

Filters:

Edge Readiness

parity (α) → edge-position transition
loading...
Loading classifier data...

Today's Markets

City Miss Rate ⓘ MAE ⓘ Date Forecast Parity α ⓘ Parity Miss α ⓘ Edge Miss ⓘ Wx ⓘ Drift ⓘ Bracket NO Ask Depth ⓘ YES Bid Edge Status
Loading...

Next-Day Markets

City Miss Rate ⓘ MAE ⓘ Date Forecast Parity α Parity Miss α ⓘ Edge Miss ⓘ Wx ⓘ Bracket NO Ask Depth ⓘ YES Ask Edge Status
Loading...

Next-Day LOW Markets

Overnight low forecast for tomorrow. Next-day only (same-day LOW has already happened by scan time). NO on the forecast bracket in high miss-rate cities.

City Miss Rate MAE Date Forecast Parity α Parity Miss α ⓘ Edge Miss ⓘ Wx ⓘ Bracket NO Ask Depth ⓘ YES Ask Edge Status
Loading...
Column Reference
Miss Rate
Bracket Miss Rate — % of days the actual high lands OUTSIDE the NWS forecast 2°F bracket. Primary classifier for NO bets.
Format: BLENDED% (14d/30d). Cities ≥ 70% are tagged HIGH MISS.
MAE
Mean Absolute Error (°F). Display only. High MAE = volatile climate.
Parity α
Alpha — superseded by Edge Miss. Assumes a fixed bracket grid (wrong for ~10 of 17 cities). Kept for comparison.
Parity Miss α
Alpha — miss rate split by odd/even forecast. Mixes opposite cases for alternating-grid cities. See Edge Miss.
Edge Miss
Miss rate split by which edge of the bracket the forecast sits on (TOP or BOTTOM). Structurally correct replacement for Parity Miss.
Warm-biased cities prefer TOP-edge days. Cold-biased prefer BOTTOM-edge. Format: TOP% (n) | BOT% (n). Today’s edge highlighted.
Wx (Weather Conviction)
Day-level forecast bust conviction. Display only — does not drive signal firing.
Shows a single confidence level on top with model spread and NWS-ensemble diff below:

Levels:
• HIGH — responsive city + trough/front in AFD + elevated AFD score. Strong bust signal → lean into NO.
• MED — some bust indicators elevated (trough/front present, high spread, or elevated AFD).
• BASE — no strong conviction either direction. Use classifier miss rate as-is.
• LOW — stable pattern, no trough/front, low AFD. NWS is more likely to be accurate → may favor YES plays.
• N/A — inverted city (LAX, SFO, SEA, MIA). The Wx column itself does not apply — AFD signal is unreliable due to marine-layer effects. Not a skip-the-trade flag — trust the pricing-layer signal as-is.

Data below the level:
• Spread (°F) — disagreement across 6 weather models (GFS, ECMWF, ICON, GEM, JMA, MeteoFrance). 3° = agree, 10°+ = major disagreement.
• Diff (°F) — ensemble mean minus NWS forecast. Large positive = models think warmer. Large negative = models think cooler.

Hover for full breakdown: AFD factor, matched keywords, tier, alerts.
Drift
Forecast change since last evening’s scan (Today table only). > 3°F drift can invalidate an evening-placed NO bet.
Depth
NO-buyer book depth. Format: top / +3¢ / +5¢ / +10¢. Green = ≥100 within 3¢, yellow = need 5¢, red = thin.
NO Ask
Kalshi price to buy NO. Green = below city cap, yellow = within 5¢. Cap = miss rate − 10pp.
Edge
Miss rate minus NO ask price. Positive = market underpricing the miss likelihood.
BUY NO
Signal fires when: HIGH MISS (≥ 70%) AND NO ask ≤ city cap. Currently dry-run only.

Strategy Details

High Miss-Rate Cities (dynamic)

Loading...

Blended 0.6×14d + 0.4×30d ≥ 70%

Historical Win Rate

75.7% NO wins (289 settlements)

Price Threshold

Per-city: blended miss rate − 10pp

Next-Day Scans

2:30 PM, 4 PM, 6 PM, 8 PM ET

Budget

$10 per signal

Mode

Dry run (accumulating data)

Live Results

REAL SIGNALS
Loading...

Signals that the Forecast NO scanner actually fired in production, settled against Kalshi’s authoritative expiration_value. Dry-run P&L only — no real trades placed yet.

Signals

—

W / L

—

Win Rate

—

Cost

—

P&L

—

ROI

—

Next-Day Entry scanned evening before

Bets

—

WR

—

P&L

—

ROI

—

Today-Of Entry scanned during settlement day

Bets

—

WR

—

P&L

—

ROI

—

By City

City Bets W/L WR P&L ROI

Recent Signals

Date City Fcst→Actual Price P&L Res

Historical Backtest

REPLAY
Loading...

Replays all historical scans through the current Forecast NO strategy rules. Each row = what the strategy would have bet on that day based on the earliest scan available. Note that today’s strategy filters (parity, bias-change, miss-rate threshold) are applied retroactively, so these numbers can disagree with the Live Results above — live reflects what actually fired on each historical day.

Settlements

-

Would Bet

-

Win Rate

-

Total Cost

-

Total P&L

-

ROI

-

By City

City Bets W/L Win Rate Cost P&L ROI

Individual Trades

Date City Fcst Actual Bracket NO Ask Cost Result P&L