Loading...
TL;DR
Sync
Pull Kalshi markets every 6 hours
Forecast
NWS + 4-model ensemble (141 members)
Score
KL-divergence weighted edge scoring
Auto-Trade
Kelly-sized orders via Kalshi API
Track
Price deltas (1h/6h/24h) + positions
Settle
Auto-settle, Brier scores tracked
1.Market Sync
A cron job syncs open Kalshi prediction markets every 6 hours. It pulls events across all categories — politics, economics, crypto, climate, culture — excluding sports (covered by our sports scanner) and ultra-short crypto contracts.
2.Price Tracking
Each market's YES/NO price, volume, and open interest are tracked over time. Price history is stored with 1h/6h/24h deltas so you can see momentum and trend direction at a glance.
3.Category Browsing
Markets are organized by Kalshi category: Politics, Economics, Climate & Weather, Culture, Tech, and more. Filter by category, sort by volume or close date, and click through to Kalshi to trade.
4.Close Date Awareness
A second sync pass specifically targets markets closing within 30 days — the most actionable time horizon. These near-expiry markets are always up to date even if they're buried in Kalshi's newest-first ordering.
5.What's Next: Edge Model
Currently this is a market browser and tracker. The next milestone is building an edge-scoring model (similar to our options and sports scanners) that generates picks with probability estimates, EV calculation, and position sizing.
1.NBA / MLB Game-Winners
Pinnacle is treated as the sharpest price in the world; the Pinnacle-vs-Kalshi gap is the primary edge signal, layered with team ratings, injuries, rest, and momentum. Auto-trades YES on Kalshi at small size (wallet ~$100). Model P&L is tracked from trade records, never from the wallet balance.
2.NFL Game-Winners, Spreads & Totals
Live since 2026-09-13 at test-bet size (3 contracts). Fair value comes from devigged Pinnacle moneylines, spreads, and totals; spreads and totals are priced against Kalshi's half-point rungs with a normal margin model, and moneyline favorites are capped at 70¢. Up to 3 picks per game (best side, best spread rung, best total rung). Settled from ESPN final scores.
3.NFL Promotion Bar
The NFL lane is a data lane until it earns size: calibration edge (actual wins minus the sum of Kalshi implied probabilities) must reach +4 pts on n ≥ 100 settled picks per market type, with flat-$1 ROI still positive after fees. Computed every Friday by the Kalshi scorecard; below the bar, sizing stays at the test bet.
1.City Weather Markets
Kalshi offers daily high-temperature contracts for 20 major US cities. Each market has YES/NO brackets (e.g. 'NYC high above 65°F'). The scanner checks all active brackets every morning.
2.NWS Forecast Comparison
The National Weather Service forecast for each city's daily high is compared against every Kalshi bracket price. When the NWS says 72°F and Kalshi prices 'above 70°F' at only 40¢, there's edge.
3.141-Member Ensemble
The scanner ingests 141 ensemble members across GFS, ECMWF, CMC, and ICON models. Edge is computed probabilistically using ensemble spread and standard deviations, not just a single NWS point forecast.
4.KL-Weighted Edge Scoring
Edges near price extremes (5¢, 95¢) are weighted exponentially higher using KL divergence. A 20% edge on a 5¢ contract scores 5.3x higher than the same edge at 50¢, reflecting informational surprise. Rain markets are also scanned using NWS precipitation probability.
5.Auto-Trade — Paused
Weather auto-trade is suspended pending model rework (the original kill-switch fired after the prior run hit 18.2% WR; that model has since been rebuilt). Pick generation is temporarily offline while the rebuild is completed; execution resumes only after the rebuilt model clears the unpause gate (Brier ≤ 0.30 over ≥ 20 settled post-cohort picks) plus a manual owner checkpoint. NBA/MLB game-winner and NFL game-winner/spread/total auto-trade via Kalshi remains live (wallet ~$100; model P&L is tracked from trade records, not the wallet balance).
Prediction Market Sources
Kalshi API
Prediction markets, prices, volume, trade execution
Weather Scanner Sources
Kalshi API
Temperature + rain contracts, market prices, settlement
NWS API
Point forecasts, hourly temperatures, precipitation probability
GFS / ECMWF
Ensemble model data, standard deviations for probabilistic edge
Important
Other Sectors