World Cup Odds: Prediction Markets vs Sportsbooks

The short answer

There are two places to find “odds” on the 2026 World Cup, and they speak different languages. In early June 2026, a sportsbook listed France at +450; a prediction market had France near 17 cents (the dated snapshots are below, and both move daily). Both encode a chance of France winning — but only one of them reads directly as a probability, and the gap between them is mostly about a hidden cost called the margin.

Translate the sportsbook line and France at +450 implies 18.2%. The prediction markets put France near 17.1%. Close — but the sportsbook number is inflated, and the reason is structural: a sportsbook bakes a large margin into its odds, while a prediction-market price carries only a sliver of one. Using a real 48-team board, this article shows exactly how big that difference is, how to strip it out, and why — if what you want is an honest probability — the prediction-market price is the more direct read. For the deeper structural differences between the two business models, this pairs with our explainer on DraftKings Predictions vs traditional sportsbooks.

Risk note: Event contracts and sports bets both carry a real risk of loss, including total loss of stake. All prices here are dated snapshots that change constantly. Nothing in this article is financial or betting advice.

Two machines that quote the same event differently

The odds look different because the businesses behind them are different.

A sportsbook is your counterparty. When it posts that France line, it is taking the other side of your bet and managing its risk across all the action. Its revenue is the vig — a margin built into the odds so that, summed across every outcome, the implied probabilities exceed 100%. You are betting against the house, and the house prices in its edge.

A prediction market — Kalshi or Polymarket — is an exchange. You trade a standardized yes/no contract against other participants; the platform only matches you and takes a small explicit fee, holding no side of the outcome. The price, in cents, reads straight off as the crowd’s implied probability. We cover that mechanic in depth in how World Cup prediction markets work and the implied-probability guide; the structural contrast — house vs exchange, vig vs fee — is laid out in our sportsbook comparison. Here we focus on the one thing a bettor feels most directly: how much margin is in the number.

The same favorite, two prices

Start with the headline contract. As of June 1, 2026, BetMGM (via Yahoo Sports) listed the co-favorites France and Spain at +450. Our synchronized prediction-market snapshot the next morning (June 2, 10:09 UTC, from each platform’s own API) put France at 17.1% on both Kalshi and Polymarket.

Convert the sportsbook line with the standard formula — implied probability = 100 / (odds + 100):

France +450 → 100 / 550 = 18.2%

So the raw sportsbook number (18.2%) sits above the prediction-market number (17.1%). That is not the book thinking France is more likely; it is the margin talking. To see how much, you have to look at the whole board.

The overround gap — the heart of it

Add up the implied probabilities of every team and you measure the total margin. We did this for both worlds.

On the sportsbook side, BetMGM’s full 48-team board summed to about 129.6% — a roughly 30-point overround. On the prediction-market side, our snapshot summed to about 102.8% on Polymarket and 104.7% on Kalshi — a few points.

Sportsbook odds: BetMGM via Yahoo Sports, June 1, 2026. Prediction-market snapshot: Kalshi and Polymarket APIs, June 2, 2026, 10:09 UTC. All odds and prices are live and change constantly.

That is the practical headline: a traditional outright (futures) market carries a margin several times larger than a prediction market’s. Futures markets are among the highest-margin bets a sportsbook offers precisely because they spread a cut across dozens of teams. The exchange, charging an explicit fee instead, leaves the price almost clean.

De-vigging: making the two comparable

To compare a sportsbook line fairly to a prediction-market price, you have to remove the margin — “de-vig” the odds. The standard method is proportional: divide each team’s raw implied probability by the board’s total.

France de-vigged ≈ 18.2% ÷ 1.296 ≈ 14.0%

So once you strip BetMGM’s ~30-point margin, its France lands near 14% — now below the prediction markets’ ~17%. The two worlds do not perfectly agree, and that is honest to report: they are different crowds with different costs, and the de-vig method itself is approximate.

One important caveat keeps us from over-reading that 14%. Betting margin is not spread evenly: the well-documented favorite-longshot bias shows longshots are overpriced and favorites underpriced, so a simple proportional de-vig understates favorites (Snowberg & Wolfers, NBER 2010). BetMGM’s “true” France is probably somewhat above 14%, but pinning it exactly requires a margin model, not a one-line division. The robust, method-independent point is the one in the table: the sportsbook number needs heavy cleaning; the prediction-market number barely needs any.

Run the same check across the board and the pattern holds. England at +650 implies 13.3% raw and de-vigs to roughly 10.3%, against about 11.1% on the prediction markets — within a point or two once the margin is gone. That is the real lesson: the prediction-market price was already telling you the contenders’ probabilities, no arithmetic required, while the sportsbook number only revealed them after you stripped out the vig.

What the margin actually costs you

The overround is the book’s built-in edge. The ~30-point figure is the total margin across all 48 teams, not the hold on any single bet — the edge you concede backing one team is smaller than that headline, but it is real, and the favorite-longshot bias makes it steeper on the deep underdogs than on the favorites. A prediction market swaps that embedded margin for a transparent, usually small per-contract fee, and leaves the quoted price close to fair. It also lets you sell before the final: because contracts trade live, you can exit to lock in a gain or cut a loss, whereas a sportsbook futures ticket generally holds you until the tournament ends, aside from limited cash-out offers. On a month-long bet, that flexibility is its own kind of value.

Why this matters for reading probability

If your goal is to know what the market thinks the chance is, the implications are concrete:

• The prediction-market price is the more direct read. France at 17.1% in our June 2 snapshot is already within a point or two of a clean probability. The sportsbook’s 18.2% is a probability wearing a thick margin coat you have to take off first.
• Sportsbook longshots are the worst value. On a 48-team board the margin piles onto the underdogs. A 1,000-1 team is not a clean 0.1% — between the vig and the favorite-longshot bias, it is materially overpriced. (Prediction-market longshots are overpriced too, just less so.)
• Neither is a forecast. Both numbers are estimates that move on the draw, injuries, and results. A cleaner probability is still a probability, not a promise.

This is the on-ramp a lot of sports bettors take into prediction markets: once you have de-vigged a few sportsbook boards by hand, a venue that just shows you the probability — and charges a small fee instead of a fat margin — starts to look like the more honest instrument. It is not magic, and it is not free, but the number means what it says.

Which should you use?

• A sportsbook if you want familiarity, promotions, and the full sports-betting feature set, and you are comfortable that the posted odds include a sizable margin you may never bother to remove.
• A prediction market if you want the price to read as a probability, a small explicit fee instead of an embedded vig, and the ability to sell a position before the tournament ends. The trade-off is a newer interface and a learning curve.

For most probability-minded readers the deciding factor is exactly the one this article measured: on the World Cup, the exchange hands you a number that is already almost clean, while the sportsbook hands you one you have to scrub before it tells the truth.

Bottom line

Sportsbook odds and prediction-market prices both answer “how likely is France?” — but the sportsbook answer is inflated by a roughly 30-point margin on the 48-team board, while the prediction-market answer carries only a few points. De-vig the sportsbook and the two move toward each other; skip that step and you are reading the house’s margin as if it were probability.

Reading a price as an honest probability — and knowing exactly how much to clean off first — is the discipline MispriceHQ is built on. To be clear and honest, as always: our machine-learning fair-value engine is in development, with no live track record and no markets resolved; we will share its methodology and performance only when it is validated. In the meantime: see how the World Cup winner market works and why Kalshi and Polymarket price the same teams differently.

Reminder: Sports bets and event contracts carry a real risk of loss. The odds and prices here are dated snapshots from June 1-2, 2026 and will have changed. This is research and education, not financial or betting advice.