Agentic research and trade-recommendation platform for prediction markets. Kalshi first, venue-agnostic by design.
Week 3 Sunday, forty-four days from the midterms, and the tropics are asleep. The exchange is pricing a Fed that hikes rather than cuts, gas above $4.50, a Florida ballot with very little suspense at the top, and a Dolphins season that is already being written off. Below is what the market believes right now, with our read on each number.
Game markets are the deepest thing on the exchange today: the Eagles–Titans contract alone has traded 5.9M contracts. Prices are live during play, so the afternoon numbers already reflect the score.
Miami opened as a road underdog and the in-game price has collapsed to a coin's edge. San Francisco at 90¢ is priced like a game that is effectively over.
A genuine toss-up in Denver, and Jacksonville money keeps arriving: the Jags side has traded nearly twice the Broncos side.
Tampa Bay was 79¢ pregame. Cleveland has made it a game, which is exactly the kind of swing a live book lets you see and a pregame line hides.
The most-traded contract on the exchange today. Philadelphia is still favored but has bled eight points of probability since kickoff.
Kansas City is a clear but not overwhelming home favorite. Next week they travel to Miami and the market already has them at 84¢.
Jacksonville is the AFC South story of the year: the market gives them a 59% shot at 10+ wins and a 50% shot at 11+.
Rams 13¢, Bills 12¢, Ravens 9¢, 49ers 8¢, Chiefs and Seahawks 7¢. No team above 13% in late September is a wide-open year, not a Rams year.
Allen is alone at the top. Caleb Williams and Joe Burrow are next at 8¢; the market has moved off Williams (12¢ → 8¢) after Week 2.
Read it the other way: a 75% chance Miami finishes 4–13 or worse. Three-plus wins is only 59¢. The rebuild is priced in.
The macro board is unusually one-directional. Cut markets are dead for the year, inflation prints are being priced above 3.5%, and energy is the reason.
Hike 56¢, hold 45¢, any cut ≈ 1¢. December leans harder toward a hike at 66¢. The fed-funds market has the rate above 4.00% after October at 61¢.
Effectively settled. Anyone who bought 'two cuts' in January is holding a 0¢ contract.
Above 3.6% is a coin flip at 46¢. The month-over-month market has a 60% chance of a 0.5%+ print, which is where the hike case comes from.
The weekly market already has the national average locked above $4.46. Florida is running cheaper: the state contract settles around $4.26–4.29 this week.
Crude is in the mid-to-high $90s with a 5% tail above $110 by Friday. That tail is what the September gas and CPI markets are watching.
Labor is soft, not broken: above 4.2% is only 11¢ and payrolls above +100k is a 41–57¢ spread. Recession-in-2026 trades at 5¢ on 3.5M contracts.
Seven percent is the line the market is arguing about: above 7.08% is 9¢. Housing is absorbing the rate path the Fed markets describe.
Equities are shrugging off the hike talk. 8,200 is 22¢. Bitcoin's year-end distribution centers on $75–85k, and 'BTC beats gold in 2026' is only 37¢.
National control markets are close; the Florida statewide races are not. The October 1 shutdown contract, which traded 859k contracts, has resolved itself to a near-certain no.
A continuing resolution is priced as done. The 'how many shutdowns in 2026' market still sits at two (64¢), so the fight is expected to return after the election.
The seat-count distribution puts most mass at 51–53 for Democrats, which is a majority. Below 48 is a 2¢ tail. Democrats are 65¢ to hold more governorships after November.
David Jolly at 22¢ has been the more active side this week (13.8k contracts vs 1k) without moving the price. The vote-share market has Donalds at 50%+ at 71¢.
Angie Nixon trades at 12¢. Compare the 2028 regular Florida Senate market, where Republicans are only 80–85¢: the market thinks the state is red, not unlosable.
The most interesting Florida contract on the board. A 40¢ price on a measure that needs a 60% supermajority means the market expects majority support but doubts the threshold.
Below 35% is 26¢ and below 33% is 10¢. The upside market has 'above 45%' at just 7¢, down from 16¢. The band is narrow and drifting down.
Vance leads at 45¢; Rubio is a clear second, DeSantis is 4¢, and a Trump family member is a rounding error. Rubio's own 'announces by Oct 2027' contract is a coin flip.
The nominee side is crowded: Ossoff 17¢, Ocasio-Cortez 16¢, Newsom 14¢, Harris 8¢. Nobody above 20¢ two years out is the normal shape, not a signal.
The peak of hurricane season has come and gone with the Atlantic market pricing almost nothing. That is a rare and, for anyone on the coast, comfortable set of numbers.
Flip it: a 60% chance the season ends with zero Category 3+ storms. 'More than 1' is 23¢. Reinsurance desks will remember this year.
A normal season is seven. The market is saying four or fewer, with 'more than 6' at 6¢. The Eastern Pacific has been the active basin instead.
Before Oct 15 is 76¢. Something is expected, but the Category 5 market has 'before Dec 1' at only 15¢, so nothing the market believes will be historic.
The daily temperature ladders are the busiest weather markets on the exchange; today's Miami set traded over 130k contracts. Tomorrow's board leans 88–91°.
Above 7 inches is 80¢ and above 13 is 17¢. A wet month by rainfall, a dry one by storms: exactly what a no-hurricane September looks like.
Three Miami-area teams sit at or near the top of their championship boards, which is not something the market has said in a while. Elsewhere: Dodgers, 76ers, Texas.
The favorite, ahead of Colorado (10¢) and Carolina (9¢). Their points market has 100+ at 74¢ before a puck has dropped.
Vancouver and Nashville are 23¢ each, but Miami has traded more than twice the contracts of any other club. They visit Columbus on Sep 27 at 43–54¢.
Fourth on the board behind Texas (15¢), Ohio State (12¢) and Ole Miss (7¢). The regular-season market has the Hurricanes at 11+ wins at 65¢; Clemson comes to town Oct 3.
Trinidad Chambliss (23¢, up from 11¢) and Darian Mensah (21¢) are the frontrunners after this week; Toney has slipped but leads the board in total contracts traded.
50+ is 37¢. The title market has Miami at 4–5¢, second only to Philadelphia at 10¢, which says more about how flat the NBA board is than about the Heat.
Sliding late. Up the road, Tampa Bay's 95-win contract is at 98¢ and the Rays are a live World Series contender behind the Dodgers (31¢) and Brewers (17¢).
Given the three boards above, 44¢ looks cheap. Two-plus titles is 14¢. This is the kind of cross-market inconsistency the scanner is built to flag.
Compiled from the Kalshi Trade API on Sep 20, 2026 · 3 PM ET across open, non-combo markets. Volume figures are lifetime contracts. In-game prices move by the minute; the linked market pages are live. Research and analytics, not investment advice.
Every run is a background job with hard caps. Every output is structured and stored before the UI ever sees it.
Find markets worth attention across all open markets
scheduled; delayed for free tier, fresh for paid tiers
Form an independent probability for one market
on demand, or automatically for paid watchlists
Turn an analyst result into a recommendation
Kelly-derived sizing, capped
Abstaining is a first-class output and the abstain rate is published, never hidden. A model that only ever says YES or NO is guessing.
Model probability sits above the market price by more than the confidence band. Entry, exit, and invalidation stated.
The mirror case: the market is over-pricing the outcome. Same structure, opposite side.
Fairly priced, or too uncertain to call. Reported as a real answer and counted in the public abstain rate.
Conditional entry with a stated trigger price. The thesis holds; the price doesn’t, yet.
Users act on recommendations themselves. The app deep-links to the market on the venue. The provider interface excludes execution methods entirely so the boundary is structural, not a config flag.
One MarketProvider interface, one normalized market model, one adapter per venue. Nothing outside the adapter may import venue-specific types.
Credibility and marketing; built before the first user. Every recommendation is logged at the price it was made, then graded on resolution.
Calibration, not vibes. A proper scoring rule that punishes confident misses.
How often the system declined to call it. Published because it’s the honest number.
What a unit stake at the recommended price would have returned, split by side and venue.
One normalized market model. Nothing outside an adapter may import venue-specific types, so the second venue is an adapter, not a rewrite.
REST v2 plus WebSocket; RSA-signed auth for authenticated endpoints
series -> event -> market hierarchy; prices in cents
Gamma API for metadata, CLOB API for prices
outcome tokens priced in USDC; first target for cross-venue checks
Metered by credits so per-action cost can be tuned without changing the product. The scanner and track record are shared computation and cheap to serve to everyone.
Payments: stablecoins only. No custody of funds or keys on our side in v1. Auth via clerk.
We took the name, not the magic. Every reading here comes with a probability, a confidence band, the rules it rests on, and the option to say nothing. The mark is a ring drawn five-sixths of the way around: it never closes, because the system never claims certainty.
Analysis and recommendation only. There is no order path to flip on.
Abstain and wait are answers. The abstain rate is public.
Every call is logged at the price it was made and graded on resolution.
The admin dashboard tracks milestones, open decisions, risks, and every spec document in one place.