Sportsbook vs Prediction Market Edge Finder — Kalshi & Polymarket

The only tool that cross-references sharp sportsbook consensus odds with prediction market prices for the same games — and flags where Kalshi or Polymarket is mispriced relative to the books.

Sportsbooks price sports better than anyone; prediction markets often lag, especially on lower-volume games. This Actor turns that gap into a ranked, quantified signal: it fetches moneylines for upcoming NFL, NBA, MLB, NHL and Premier League games, strips the vig to get fair probabilities, finds the matching winner-markets on Kalshi and Polymarket, and reports every discrepancy above your threshold — Kalshi taker fees included.

How it works

1. Book odds (keyless) — moneylines for upcoming games from a public feed (DraftKings lines via ESPN). Optionally plug in a free the-odds-api.com key to upgrade to a multi-book consensus average.
2. De-vig — implied probabilities are normalized to sum to 1 (multiplicative devig), producing the books' fair probability for each side (3-way with draw for soccer).
3. Prediction markets — live two-sided quotes from Kalshi (parlays excluded) and binary markets from Polymarket.
4. Matching — a game matches a market when both team names appear in the market title (city/nickname/abbreviation aliases handled) and the market closes within ±1 day of kickoff. Every row carries match_confidence, matched_terms and a warning so you can verify.
5. Edge math — edge_pct = (book fair prob − market YES cost) × 100, with Kalshi taker fees baked into the cost. Positive edge = the prediction market underprices that side.

Output fields

Field	Description
sport, game, kickoff	Which game (e.g. Celtics @ Lakers)
book_provider, book_ml_home, book_ml_away	Where the lines came from
book_fair_prob_pct	Vig-free book probability for the flagged side
pm_platform, pm_market, pm_url	The matched prediction market
pm_side_team, pm_prob_pct	Which team YES backs, and its all-in cost (fees included)
edge_pct, direction	Signed edge and a plain-English signal
pm_volume_24h, pm_liquidity	Judge executability before acting
match_confidence, matched_terms, warning	Transparent matching — verify before trading

A full report (games scanned, market counts, errors) is saved to the key-value store as REPORT.

Use cases

• +EV hunting on Kalshi and Polymarket against sharper prices
• Research — how efficient are prediction markets vs Vegas, by sport and volume?
• Agent pipelines — MCP-ready output schema; feed edges into alerting or execution bots
• Pair with my other Actors: Polymarket × Kalshi Arbitrage Scanner for risk-free cross-platform arbs and Kalshi Markets & Odds API for the raw normalized data.

Honest limitations

• This is +EV signal, not arbitrage. An edge vs the books is probabilistic — books and prediction markets can both be wrong. Expect variance.
• Matching is heuristic. Team-name matching with alias tables is robust but not perfect — always open both links and verify the market resolves on exactly this game before trading. That's why every row shows its matching evidence.
• Odds availability follows the calendar. Off-season leagues return nothing; lines appear as games approach. Schedule runs close to game days.
• Book lines move; prediction-market asks are order-book snapshots. Freshness matters — run on a schedule.

Not financial advice. Sports trading involves risk.

Pricing

Pay-per-event: a flat fee per scan plus a small fee per edge found. No edges, almost nothing charged.