Polymarket + Kalshi Arbitrage Opportunity Radar

Find equivalent contracts across Polymarket and Kalshi, reject false arbitrage before it reaches the economics layer, and estimate whether a cross-venue trade is actually executable after order-book depth and fees.

The Actor is designed for research, monitoring, agent workflows, and automated opportunity triage. It does not assume that similar titles represent the same bet.

What you get

Each validated row can include:

• canonical cross-platform event identity
• contract type and scope
• match confidence
• resolution-risk score and reasons
• direct Polymarket and Kalshi market data
• executable contracts from live order-book depth
• average and worst fill prices
• estimated platform fees
• settlement payout
• estimated net profit
• estimated net edge
• ACTIONABLE, WATCH, or REJECT
• NEW, CHANGED, or UNCHANGED for monitoring runs

A separate run summary reports discovery counts, canonical coverage, shared event keys, semantic-gate counts, source diagnostics, and signal totals.

Safety-first matching pipeline

Polymarket + Kalshi discovery
          ↓
platform normalization
          ↓
canonical event fingerprint
          ↓
contract type + contract scope
          ↓
resolution-equivalence gate
          ↓
ONLY validated pairs
          ↓
live order-book depth
          ↓
fee-aware execution economics
          ↓
ACTIONABLE / WATCH / REJECT

The Actor intentionally performs semantic validation before arbitrage math.

Supported matching

Sports

Sports matching normalizes league, teams, scheduled date, market type, outcome side, and contract scope.

Cross-venue execution economics are precision-first. Only:

FULL_GAME moneyline / winner ↔ FULL_GAME moneyline / winner

can pass the execution gate.

The Actor rejects derivatives such as:

• first-half or quarter moneylines
• spreads
• totals
• inning markets
• exact winning margins
• longest field goal
• scoring props
• safety / overtime props
• player props

even when they share the same underlying game.

Macro

Canonical matching supports market families such as:

• Fed meeting decisions
• CPI
• unemployment
• GDP
• payrolls

Metric, period, comparator, threshold, and contract semantics are checked before economics.

Crypto and weather

The Actor normalizes asset/location, metric, target date or period, comparator, and threshold before pairing contracts. Unsupported or non-equivalent structures are rejected conservatively.

Signals

ACTIONABLE

A research signal that passes the configured semantic, resolution-risk, fee, depth, and net-edge gates.

It is not an execution guarantee.

WATCH

The pair is semantically valid, but current execution economics or another configured threshold is not strong enough for ACTIONABLE.

REJECT

The pair fails a semantic, contract-scope, resolution, or hard safety gate. Rejected rows are omitted by default and can be included for audits.

Quick start — NFL opportunity scan

{
  "searchQueries": ["NFL"],
  "verticals": ["SPORTS"],
  "maxMarketsPerPlatform": 1000,
  "maxMatches": 10,
  "assumedTradeSize": 25,
  "minNetEdgePct": 1,
  "minExecutableContracts": 5,
  "candidateMatchConfidence": 0.72,
  "actionableMatchConfidence": 0.9,
  "maxResolutionRiskScore": 45,
  "polymarketFeeMode": "AUTO",
  "requireKnownFeesForActionable": true,
  "includeRejected": false,
  "compareWithPreviousRun": false
}

Contract-equivalence audit

To inspect why same-event contracts were rejected:

{
  "searchQueries": ["Yankees"],
  "verticals": ["SPORTS"],
  "maxMatches": 20,
  "includeRejected": true,
  "compareWithPreviousRun": false
}

Use the Semantic & resolution risk dataset view to inspect market type, scope, match confidence, and risk reasons.

Monitoring

Enable:

{
  "compareWithPreviousRun": true,
  "historyKey": "my-strategy-watch"
}

Runs with the same history key can label matched opportunities as NEW, CHANGED, or UNCHANGED.

This is useful for Apify schedules, webhooks, agents, and recurring market scans.

Fee-aware economics

The execution layer walks multiple order-book levels and estimates actual fillability rather than assuming unlimited top-of-book size.

For each strategy it can report:

• executable contracts
• average ask per leg
• worst ask per leg
• gross cost
• estimated fees
• total cost
• settlement payout
• estimated net profit
• estimated net edge

polymarketFeeMode=AUTO uses the supported category fee treatment when it can be established. Unknown fee treatment remains conservative when requireKnownFeesForActionable=true.

Dataset views

The default dataset includes three useful views:

• Opportunity overview — signals, markets, match/risk scores, edge, and executable size
• Execution economics — strategy, cost, fees, payout, profit, and net edge
• Semantic & resolution risk — event identity, contract types/scopes, confidence, and rejection reasons

Cost controls

The Actor uses pay-per-event Store pricing.

Current published pricing is intentionally simple:

• Actor start event
• per-result dataset event

maxMatches is the main billable-output safety limit. Lower it for exploratory runs.

Agent and automation use

The Actor is designed for structured API/agent workflows:

• JSON input
• structured dataset rows
• deterministic signal fields
• run diagnostics in OUTPUT
• persistent monitoring keys
• limited Actor permissions
• no Standby dependency

Important limitations

• Cross-venue rules can differ even when event identity is exact.
• Cancellation/postponement rules can create material settlement risk.
• Order books can change between data capture and execution.
• Platform access, account eligibility, balances, and order fills are external to this Actor.
• Public platform schemas and fee schedules can change and require maintenance.
• A result labeled ACTIONABLE is a research classification, not financial advice or an execution guarantee.

Version

Store-ready source package: 2.8.1

Validated matching engine: 2.8.0 / contract-semantics-v7