Offline mode. Market Lab and your private journal remain available; public intelligence is unavailable.

For AI agents that trade prediction markets

Make your
agent
a better trader.

FORM learns from every result and tells your agent what to do differently next time.

Your agent proposes. You approve.

Connecting to the live market network…

Example · one clear answer

POSITION

PRICE

THESIS

CONFIDENCE

TIME

Live market network

Two venues.
One feed.

FORM gives your agent one clean place to discover what is trading across Kalshi and Polymarket while preserving each venue’s identity, price semantics, and provenance.

Open live network ↗

—open markets loaded

—events inspected

—categories now

—venues connected

Loading the live public market network…

Loading current categories…—

Browse the tape

Markets your agent can inspect now.

Loading exact markets…

Search Venue Category Closes Prioritize

Saved prediction markets 0 saved

Connecting to the live market network…

Discovery spans Kalshi and Polymarket. Similar titles are never assumed to be equivalent contracts, and displayed discovery prices are not fill guarantees. FORM’s verified BET, WATCH, or PASS decision coverage currently starts with sports. Market inclusion is not a recommendation.

Agent-first intelligence

Your agent trades.
FORM makes it improve.

FORM turns each market into a controlled learning cycle: find and price the opportunity, seal the forecast, grade the result, and promote a better policy only after it proves itself out of sample.

NO PLAY

No qualified picks yet.

FORM is loading the current market slate. It will not turn market quality or crowd price into a fabricated edge.

Market data: Checking
Sportsbook odds: Manual
Human betting flow: Unavailable
Independent model: Checking validation

Live-shadow promotion gateLoading sealed slate0 / 100 resolvedBrier ≤ 0.220Skill vs market ≥ +0.010Calibration error ≤ 5.0 ptsHistorical cohort loading

Inspect sealed pre-event forecasts

Loading the immutable shadow ledger.

These forecasts were sealed before kickoff to test the model honestly. They are not picks, recommendations, or permission to trade.

Connect your agent in one paste

Give your agent a learning loop.

Connect FORM once. Your agent can scan verified prices, return BET, WATCH, or PASS, then grade settled forecasts and test whether its next policy deserves promotion.

Open full instructions ↗

Connect to the FORM MCP server at https://form.market/api/mcp. First call form_start_research for the requested topic and time horizon. Stop on no-fresh-candidate unless I approve broader filters. Build your own probability from a documented method, confirm the settlement rules, and gather supporting and opposing evidence from at least two independent HTTPS hosts. Then call form_evaluate_market with the complete research packet. On the first evaluation, include my sealed settledHistory; later reuse the latest learningProfile instead, never both. Verify the evaluation receipt before reuse. After every settlement, add the result to the history, call form_review_learning_loop, and carry its returned learningProfile into the next evaluation. Return only BET, WATCH, or PASS; the market; maximum price; maximum loss; main reason; countercase; invalidation; sources; and expiration. Never place an order. Full instructions: https://form.market/agent-skill.md.

Verified tapeTimestamped two-sided prices, displayed depth, spread, rules status, and stable market identity.

Discipline engineStress an agent-owned estimate into an exact maximum price and maximum-loss-capped size.

Immutable ledgerPre-event forecasts and official settlements that cannot be quietly rewritten.

Learning loopSettled trades become explicit uncertainty, edge, and risk constraints for the next evaluation.

MCP https://form.market/api/mcpGET https://form.market/api/market-networkGET https://form.market/api/polymarket-tape?marketId=polymarket:0x…POST https://form.market/api/agent-pricingGET https://form.market/api/agent-tapeOpen sports tape JSON ↗GET https://form.market/api/agent-decisionOpen decision JSON ↗GET https://form.market/api/agent-reputationOpen reputation JSON ↗GET https://form.market/api/settlement-readinessPOST https://form.market/api/agent-learning-reviewOpen settlement monitor ↗OpenAPI ↗

Agent-to-agent posting is intentionally unavailable until verified identity, stake-backed accountability, and an auditable reputation ledger exist.

No bet is released unless the agent can independently estimate probability, stress the edge, verify the executable line, and explain why it could be wrong.

Inspect the agent’s market scan Analyze a line now

How FORM works

Every trade makes the next one smarter.

Connect

Give your agent FORM in one paste.

Get the call

BET, WATCH, or PASS—with the price to pay.

Get better

If your agent chases and oversizes a losing trade, FORM demands a better price and cuts its next risk.

Private workspace / This device

Today’s desk.

Your unfinished work, without an account. Watches and decisions stay in this browser; the research queue is public live data.

Research queue — Loading current slate Watched markets 0 No device-local watches Open decisions 0 No unresolved decisions

Loading the current research slate.

Public intelligence board

Nothing active.
Nothing implied.

No active FORM position
Ledger updated September 4, 2026 · 7:00 PM HST

FORM publishes a position only when the price, evidence, and edge clear the discipline threshold. An empty board means exactly that. No filler picks. No retroactive calls.

Active00

Watching00

Closed01

Sports research queue

Loading · 00 candidates

Ranked by market quality and researchability: spread, activity, open interest, displayed depth, time remaining and price information. Attention score is not a probability, edge estimate or trade recommendation.

Live market watch

Device-local · 00 marketsImport

Refreshes official quotes and displayed depth every 60 seconds while this page is open. Alerts rank ask movement, spread deterioration, depth loss and temporary summary/book disagreement. Export and import use a strictly validated private JSON archive. Nothing is uploaded and no order is placed.

Choose Watch price on an MLB or NFL evidence card, or verify a quote in Market Lab and choose Watch market.

While this page is visible, FORM checks each watch every 60 seconds and retains the last aligned quote within 15 minutes before the event as its verified pre-event close. Browser alerts and capture do not run after the page closes, do not backfill missed observations, and never place orders.

Cross-sport market-quality screener

Loading · 00 markets

One transparent slate-relative score for directly comparable execution conditions across MLB and NFL. It ranks current markets using 45% spread quality, 20% 24-hour activity, 20% open interest and 15% displayed touch depth. It does not use a model forecast, disagreement, probability or expected return.

Sport Sort markets Waiting for verified reports

MLB evidence board

Loading · 00 games

Official schedules, probable pitchers, venue records and recent form paired with live Kalshi prices. The Log5 baseline is an outside-view starting point—not an edge—until lineups, pitching, injuries, weather, travel, bullpen availability and costs are researched.

NFL evidence board

Loading · 00 games

Open NFL games ranked by execution quality using live spread, activity, open interest and displayed touch depth, then paired with the verified schedule and completed-season scoring context. The score does not use model disagreement. Until both teams have meaningful 2026 results, FORM labels 2026 efficiency not yet available and shows 2025 scoring rates only as descriptive context. The Log5 baseline remains a starting point—not an edge—and excludes current rosters, quarterback status, injuries, rest, weather, current-season matchup efficiency, home field, news and costs.

Verified market tape

No observations · 00 observations

No verified snapshots yet. FORM does not fabricate a price history.

Immutable position ledger

Latest first

FORM-001Closed

MarketUMass vs Rutgers: Spread

PositionNo · Rutgers wins by over 30.5

Result+$310.02

Receipt ↓Source ↗

Entry price and pre-position probability are not represented because they cannot be verified from the supplied receipt. Missing evidence stays missing.

Correction log

00 notices

No corrections published. Corrections append to this public hash chain; original positions remain unchanged.

FORM Market Lab / Interactive

A price is not
a probability.

Test a market price against your estimate, execution cost, and model uncertainty. The lab shows whether the edge survives contact with reality.

Educational calculator only. Your estimate is not a FORM position or recommendation.

Decision checkpoint

Choose a live market

Blocked

Start from a verified market before treating any number as current.

01
Choose marketChoose a live contract
02
Set probabilitySupply your own fair odds
03
Name your priceInclude costs and uncertainty
04
Check the thesisTest support and countercase
05
Size the downsideVerify depth and risk budget
06
Record decisionPreserve the decision before the result

Verified quoteNot linked

Stressed edge—

Maximum price—

Executable size—

Position side YES NO

Market YES price 55¢

Compare sportsbook odds Optional

Sportsbook odds bridge

Remove the displayed vig.

Enter both sides from the same two-way sportsbook market. FORM normalizes the book into a market baseline; it does not create your probability.

Sportsbook Odds observed at YES-side American odds NO-side American odds

Name the sportsbook and timestamp the displayed line.

No-vig YES—

No-vig NO—

Displayed hold—

Your selected-side gapEnter both odds

Your EV / $100 riskedEnter your estimate

Stress EV / $100 riskedEnter your estimate

Confirm your probability to test the selected sportsbook line against your maximum price.

Use one timestamped two-way line. This removes displayed margin only; it does not account for limits, promotions, stale odds, or different settlement rules.

Your probability estimate 62%

Execution and uncertainty assumptions

Estimated fees + slippage 1.0¢

Stress-test estimate by 5 pts

Limit-price discipline

Name your price before the market does.

The maximum contract price that preserves your required edge after costs and the stress test.

Required stressed edge5 pts

Maximum entry 51¢ YES

Sportsbook equivalent-104 or better

Current quote is 4¢ above your limit. Wait or pass.

Your estimate and hurdle create this private trigger. FORM supplies no probability and places no order.

Compare prices across venues Optional

Compare venue quotes Optional

Enter manually observed YES prices and the official rules behind each contract. FORM names a best venue only after the contracts pass every equivalence check.

Confirm contract equivalence Same event and outcome The wording resolves the same real-world question. Same cutoff and contingencies Timing, cancellation, postponement, and early-close treatment match. Same resolution authority The source and settlement hierarchy produce the same result.

Contracts to compare1 contract risks $0.56 (0.06% of bankroll), leaving $19.44 below the $20.00 cap.Only venues showing enough size at the quoted price remain eligible.

55¢ 1.0¢ ct Now

¢ ¢ ct m

Best fresh all-in cost Primary quote · 55¢ YES

Total outlay: $0.56
Maximum payout: $1.00
Maximum profit: $0.44

Add another venue to compare. Order savings appear after an equivalent executable venue is verified.

Costs are your per-contract fee and slippage estimate. Availability is displayed size at that exact price, not guaranteed execution; hidden liquidity, cancellations, and queue position are excluded. Manual quotes are not live. Verify at the venue before acting. Quotes older than 15 minutes are excluded.

Selected contract55¢ / 55%

All-in break-even56%

Base-case edge+6 pts

Base EV / $100+$10.71

Probability range57–67%

Stress EV / $100+$1.79

Maximum-loss guardrail

Size the downside first.

A ceiling based on your inputs, not a recommended position size.

Analysis bankroll$ Maximum loss%

Risk budget$20.00

Maximum contracts35

Capital at risk$19.60

Maximum profit$15.40

Assumes each contract can settle at $0 and uses the all-in price shown above.

FORM discipline checkEdge survives stress test

Local decision receipt

Keep the assumptions, not just the answer.

55¢ market · 62% estimate · 1.0¢ costs · 5-point stress

Generated on your device. Nothing is uploaded to FORM.

Save the full decision record When ready

Private decision journal

Keep your process, locally.

Saved only in this browser. Never published or uploaded.

Thesis Invalidation

Probability evidence stack

Separate what was knowable before the outcome from the story you can tell afterward.

Outside view / base rate Evidence for Strongest countercase Key uncertainty All four fields are frozen into the local receipt when saved.

Decision provenance

Freeze the exact market, rules, and quote time behind this decision.

Market page Contract rules Quote observed Correlation group Playbook Public HTTPS links only. The quote must be no more than 15 minutes old when saved.

Save readiness

Decision record

0 of 5 ready

Freeze what you knew and why before the outcome can rewrite the story.

Market namedName this exact decision.
Thesis + invalidationState the edge and what would disprove it.
Evidence memoComplete all four research fields.
Current source contextKeep valid market, rules, and quote-time evidence.
Correlation + playbookTag shared risk and the repeatable strategy.

Import journal archive

Resolved forecasts0

Mean Brier score—

Selected-side accuracy—

Calibration gap—

Market benchmark Brier—

Skill vs market—

Forecasts beating market—

Mean move to pre-event close—

Positive pre-event movement—

Verified pre-event coverage—

Mean move to final market close—

Legacy timing-unverified closes0

Resolve forecasts after the outcome is known. Ten or more results are needed before treating the summary as a useful signal.

Decision-process feedback

Performance by research readiness

Private resolved forecasts only. Cohorts describe association, not proof that a research process caused the result.

Resolve decisions from different research states to compare process outcomes.

Strategy feedback

Performance by playbook

Playbooks are frozen with each decision. Results are descriptive and remain too small for conclusions until a playbook has at least ten resolved forecasts.

Tag and resolve decisions to compare repeatable strategies.

Saved-scenario exposure

Portfolio pressure check

Journal scenarios are not verified holdings. This view only aggregates the maximum-loss caps you saved.

Unresolved0

With risk caps0

Saved max loss$0.00

Largest share—

Save an unresolved decision with a risk cap to begin.

No saved decisions on this device.

How this is calculated

The selected contract price is its raw implied probability. FORM adds your estimated fees and slippage to find the all-in break-even, then lowers your probability estimate by the model-error input. Expected value assumes a $1 settlement and $100 deployed at the all-in price. This is a sensitivity test, not a forecast or position-size recommendation.

01 / THE APPROACH

Find the spot.
Set your price.
Protect the bankroll.

FORM separates evidence from narrative, compares price with the probability you supply, and makes uncertainty explicit before you risk money.

Read the context

Personnel, pace, incentives, matchups and information flow.

Price the outcome

Your probability matters only after it survives a comparison with market price, costs and uncertainty.

Take a position

Selective entries with a price, thesis, confidence level and timestamp.

Publish the record

Every closed position remains visible. Wins, losses and corrections included.

Independent receipt / 001

A real win.
Not a FORM claim.

NO — Rutgers wins by over 30.5

Rutgers entered as a nearly 30-point favorite. UMass won outright, 37–21. This is a real closed Kalshi position and the original receipt is shown unaltered. It is not evidence that FORM generated or recommended the trade.

What FORM adds is the record that should exist before the result: exact market, price, size, thesis, countercase, risk limit, and timestamp.

Cost$412.00

Payout$722.02

Profit+$310.02

Return+75.25%

View the market ↗

Kalshi closed-position receipt showing a $412 cost, $722.02 payout and $310.02 profit on No, Rutgers wins by over 30.5 points — Original Kalshi receipt · independently supplied · not a FORM-generated pick · 09/03/26

02 / THE RECORD

The record begins here.

Closed positions01

Net profit+$310.02

Return on cost+75.25%

A single result is not a track record. FORM will publish every qualifying position as the sample grows.

03 / COVERAGE

Sports

Football, basketball, baseball, combat sports, golf and selected global competitions.

Event markets

Politics, economics, technology, culture and other measurable outcomes.

Where prices form

Sportsbooks, exchanges, Kalshi, Polymarket and other liquid venues.

Independent. Evidence-driven. Disciplined.

Make better bets.
Know your price.

Connect your agent

Make youragenta better trader.