Why Polymarket Sometimes Disagrees with Sportsbooks

Sportsbooks and prediction markets quote prices on the same underlying events but through structurally different mechanisms. The mechanisms produce systematic disagreements that, if persistent, are exploitable. This article explains the disagreement structure and what it means for cross-market trading on BaseCase.

Two pricing mechanisms

A sportsbook's price is a quote against the house. The book sets the line, takes the other side, and embeds a margin (vig) to cover operating cost and information disadvantage. The book actively manages risk by moving the line in response to incoming volume — sharp action moves the line toward fair value, square action gets absorbed at the book's expense. The book's price reflects its judgment of true probability after running this flow-management process across many bettors.

A prediction market's price is the result of a peer-to-peer order book. Traders post limit orders to buy or sell event contracts at specific prices; the platform matches them when bids and asks cross. There is no house balance to manage. Platform revenue comes from a flat fee per filled trade rather than from a margin embedded in the price spread. The mid-market price reflects the marginal indifference price of whichever traders happen to have unfilled orders resting at the time.

These mechanisms can produce nearly identical prices on heavily-traded liquid markets. They diverge on thin markets, on markets where the trader populations are structurally different (a sportsbook prop market vs. a Polymarket political market with the same underlying), and on markets with structural product differences (3-way moneylines on prediction markets vs. 2-way moneylines on US sportsbooks).

Where they systematically disagree

Three patterns recur:

1. Information staleness. When news breaks, the side with deeper liquidity and faster re-pricing moves first. Sportsbook lines often lead prediction-market mid-prices on sports news, because professional sportsbook flow tracks news quickly. The reverse is also true on political and macro events, where prediction-market traders are often the dominant price source and sportsbooks are the laggards.

2. Product structure. A US sportsbook moneyline prices a 2-way regulation-plus-overtime outcome. A Kalshi or Polymarket contract on the same game may price a 3-way regulation-only outcome. The probabilities are not directly comparable; they are different products that share a name. Naive cross-pricing without accounting for the structural difference produces artifacts that look like arbitrage but are not.

3. Vig vs. fee structure. A sportsbook embeds the cost of being a counterparty into the spread. A prediction market charges a fee on the trade. On a 2-way market with a 4% sportsbook vig and a 2% Kalshi taker fee, the same true probability appears on the books at different prices, and the gap is structural — it cannot be arbitraged away because the friction layer is different on each side.

How BaseCase surfaces this

The pipeline tags any leg involving Kalshi or Polymarket with is_pm = true. The flag drives two visual treatments in the Edge Finder and Arb Finder:

• A small cyan PM tag next to the platform name on each affected row.
• A cyan left-border accent on the row when no stronger signal (such as best-in-game green) overrides it.

When two PM-flagged legs combine across platforms with a sportsbook leg into a tradeable arb, the resulting opportunity surfaces in the Cross-Market Arbitrage section of the Arb Finder, gated to Sharp tier and above.

Honest current state

Polymarket integration is producing very few matched events in BaseCase's current production output. The event-matching layer that joins Polymarket contracts to sportsbook events is still under development, and most Polymarket markets do not yet have a matched sportsbook counterpart in the system. Kalshi coverage is materially better — most Kalshi sport-event contracts are matched and producing pricing comparisons — but still limited relative to the platform's full offering.

Expanding prediction-market coverage is on the roadmap. Until the matchers ship, the cross-market opportunity flow on BaseCase is biased toward Kalshi-anchored arbs and away from Polymarket-anchored ones. The PM badge surface area on rendered rows reflects this: Kalshi rows appear regularly; Polymarket rows are rare.

Caveats

Cross-platform price disagreements are not always exploitable. The most common reason: the disagreement reflects genuine product differences (resolution rules, regulation-only vs. overtime-included) rather than a pricing inefficiency. The second most common: the gap is smaller than the round-trip friction. The third: the prediction-market side has resting orders sized too thin to fill at the quoted price.

A prediction market disagreeing with a sportsbook is information. Whether it's an opportunity depends on which one is wrong, and on whether the gap survives the friction.

Further reading

• Cross-market arbitrage — when the disagreement is exploitable