What Polymarket and Kalshi Tell You About a Game
A sportsbook line and a prediction-market price answer the same question in two different languages. Learn to read both and you get something most bettors never have: a second, near-unbiased estimate of how likely a team really is to win. Here's how prediction markets differ from a book, when their prices are sharp, when they're just noise, and how we use them as a cross-check on every pick.
Two ways to price the same game
When FanDuel posts a team at −120, it's making a claim about probability (around 54.5%) and adding a margin so the book wins over time no matter who covers. Post both sides and those implied probabilities sum to more than 100%; the overflow is the vig, the house's cut.
A prediction market works differently. On Polymarket or Kalshi, a contract pays out $1 if an outcome happens and $0 if it doesn't. Traders buy and sell that contract, and the price settles where buyers and sellers agree. If a contract trades at 48 cents, the market is saying the event is about 48% likely. There's little or no margin between the two sides, so the price reads as a near-direct estimate of probability.
prediction-market price ≈ probability
No vig is the whole headline
Strip the tax and you can see what the market actually thinks. That's why a prediction-market price is so useful as a reference: it's already de-vigged for you. To compare it fairly to a sportsbook you'd normally have to remove the juice from the book's two-sided price first; the market hands you the clean number directly.
Real money, two flavors
The two venues people mean by "prediction market" settle differently. Polymarket is a real-money market that settles in crypto, with deep liquidity on the events traders care about. Kalshi is a CFTC-regulated exchange that lists event contracts and settles in US dollars; because it's regulated as a derivatives venue rather than a sportsbook, the specific contracts it offers can differ and can change over time. Both are real money, which is what makes their prices worth listening to — play-money markets drift because nobody is punished for being wrong.
The catch: liquidity
A prediction-market price is only as sharp as the money behind it. A headline game with deep volume gives you a price you can trust. A backwater matchup with a few hundred dollars of open interest gives you a number that can sit stale for hours or swing on a single trade. The price looks just as confident either way, so you have to check the volume before you believe it.
- Deep market — many traders, tight spread, price tracks reality. Treat it as a sharp signal.
- Thin market — little volume, wide spread, price lags. Treat it as barely a signal.
- No market — the event isn't listed. Common for early-round tennis, undercard fights, and weekday games. Absence of a price is not a price.
How to actually use them
For most bettors the prediction market isn't where you bet — it's where you check. Before you take a sportsbook line, glance at the market's price on the same side:
- If the market sits well below the book's implied probability, the book line may be stale in your favor.
- If the market and a sharp book agree and your book disagrees, your book is the outlier — usually a reason to pause, not pounce.
- If two independent markets (say Polymarket and Kalshi) and a sharp book all cluster on one read, that consensus is about as close to "the answer" as you'll get pre-game.
How we use them
We pull both Polymarket and Kalshi for every supported game and blend them into the model at a small weight, alongside the sportsbook line. More importantly, we use them as a consensus check: when Polymarket, Kalshi, and Pinnacle's closing price all line up against the recreational book, that agreement is a strong signal and the model leans on it. And we guard the inputs — a market below a volume floor, or one that matched the wrong game or date, is dropped so a near-empty contract never moves a pick. You can see both prices, per game, on the model page and on every individual pick page.