Before the 2026 World Cup kicked off, two authoritative systems released their respective "championship probabilities"—and their top picks differed.
The prediction markets (aggregated prices from Polymarket, Kalshi) listed France as the favorite, at approximately 17%. The Opta supercomputer listed the Euro champion Spain as the favorite, at 16.1%.
Both figures look like "probabilities." But their production methods are completely different—one is the price cleared by hundreds of millions of dollars in trading volume on the market, the other is a frequency derived from simulating the entire World Cup ten thousand times by a supercomputer.
This article does not predict who will win, nor does it evaluate which system is more accurate. It only answers one question: When you see the number "France 17%," how exactly did it come about, and to what extent is it credible?
This is the next layer after EP06—the previous article discussed how the market structure of prediction markets differs from traditional sports betting; this article explains how the probability in that price is calculated. Data as of May 31, 2026.
Act I · Probability in the Price: How the Market Produces Probability
The mechanism of prediction markets is clean: each outcome's contract is priced between 0 and 100 cents, with the price directly read as implied probability. A French contract quoted at 17 cents means the market believes France has approximately a 17% probability of winning—correct guesses pay out $1 per contract, incorrect ones pay $0.
However, prices on a single platform can be noisy. Aggregators (like DeFi Rate) use Volume-Weighted Average Price (VWAP) to aggregate quotes from multiple venues such as Kalshi, Polymarket, Polymarket US, Gemini, etc., on an hourly basis, deriving a cross-platform implied probability. As of May 30, 2026, the cumulative trading volume for World Cup champion contracts reached approximately $523 million, with the settlement date set for July 20, 2026—the day after the final on July 19.
This price does not appear out of thin air. It is the result of market makers continuously quoting bid/ask prices + traders continuously executing transactions. It's worth noting that the liquidity providers for prediction markets are all crypto-native institutional trading firms: Wintermute (annual trading volume exceeding $3.5 trillion, covering over 70 exchanges) began providing two-sided quotes for Polymarket and Kalshi in 2026; Jump Trading, Susquehanna are also active market makers.
Wintermute's OTC trading head Jake Ostrovskis summarized the market's current state in one sentence:
"Prediction markets have the demand profile of a major asset class but the liquidity profile of an early-stage one."
In other words—the credibility of that "probability" in the price depends on how much real liquidity is supporting it behind the scenes. We'll return to this in Act III.
Act II · Probability in Simulation: How Models Produce Probability
The Opta supercomputer takes a different path. It first uses team data—form, historical records, world rankings, latest international match performances—to estimate win/draw/loss probabilities for each match via Power Rankings (an Elo-derived rating algorithm). Then, it simulates the entire World Cup 10,000 times, counts how many times each team wins in the simulations, and that frequency becomes its "championship probability."
The 2026 results (stated as facts, not predictions): Spain 16.1% (also the only team with over 50% probability of reaching the quarter-finals, at 52.1%), France 13.0%, England over 10%, defending champion Argentina ranked fourth also over 10%, Portugal 7.0%, Brazil 6.6%.
There is a counterintuitive methodological detail worth highlighting here: One of the inputs for the Opta model is the odds from the betting market. This means the "market vs. model" comparison is not between two completely independent systems—the model has already partially "ingested" market information. The differences you see when comparing market prices to Opta probabilities are smaller than the divergence between two independent sources.
A timeliness note is needed: many remember the authoritative FiveThirtyEight soccer model (SPI), which stopped updating after founder Nate Silver's departure in 2023; the original website closed in September 2023, and the entire 538 was shut down by ABC in March 2025. This article treats it only as historical methodology and comparative material for the 2018 and 2022 tournaments, not as a current prediction source for 2026.
Act III · Which is More Accurate? An Honest Blank
Market or model, which is more accurate?
The honest answer is: There is no rigorous cross-tournament academic study directly comparing the Brier scores (a standard measure of prediction accuracy) of prediction markets versus Opta/538 for the 2018 and 2022 World Cups. Self-reported numbers like "90% accuracy rate" often come from the platforms themselves or non-peer-reviewed blogs and cannot be considered independent conclusions. This article explicitly states this gap, without fabricating an answer.
However, there is a commonly misstated case worth correcting. Many say "Argentina's 2022 win was a huge upset"—this is not accurate. Pre-tournament, Argentina was the second or third favorite: Opta gave 13.1% (second), bookmakers offered odds around +500 (approx. 16.7%, second). The real story isn't "an underdog win," but rather—nearly all mainstream models and markets favored Brazil, yet the second favorite Argentina won; and the only outlier that placed Argentina around 8% was precisely FiveThirtyEight. This is more precise and more telling than "upset win": so-called "authoritative probabilities" can differ by a factor of two between sources.
Price itself is not a perfect probability either. A phenomenon verified repeatedly for nearly a century is the longshot bias: in classic horse-racing markets, bettors systematically overestimate longshots and underestimate favorites—longshot horses' true win rates are lower than what the odds suggest, so betting on longshots long-term loses more (research by Snowberg and Wolfers).
The truly counterintuitive part is: this bias has not disappeared in the supposedly more rational, efficient crypto prediction markets. Multiple studies based on vast datasets from Polymarket and Kalshi have found a bias in the same direction—University College Dublin analyzed over 300,000 Kalshi contracts, finding that low-priced contracts actually materialized less often than their price-implied probability, while high-priced contracts materialized more often (i.e., longshots are still overestimated); a calibration study based on 292 million trades (arXiv preprint 2602.19520) also found that long-term contract prices are systematically compressed towards 50%, underestimating the true advantage of favorites. A microstructure preprint based on 30 billion order book events over 52 days (arXiv 2604.24366) quantified the cost on the longshot end: the bid-ask spread for the lowest probability contracts was as high as 1,300 to 1,800 basis points—an order of magnitude larger than traditional markets—rooted in market makers pricing the inventory risk of "bounded upside, asymmetric downside."
In other words: a bias recorded at racetracks a century ago still holds today in on-chain, multi-billion dollar volume markets—the "probability" in the price is less reliable the closer it gets to the longshot end.
The Ledger is Public
There is something traditional betting cannot do: Polymarket is built on Ethereum smart contracts, every trade is on-chain and auditable by anyone. The reason the two studies above were possible is precisely because researchers could directly reconstruct the direction of every trade from the on-chain transaction records—something impossible in closed-ledger traditional betting. Settlement is also on-chain: using USDC as collateral, smart contracts settle automatically, eliminating the need to trust a centralized bookmaker to hold your funds.
But transparency does not equal immunity to manipulation. Thin order books mean small markets can be moved by relatively little capital. During the tournament (June 11 to July 19), contract prices for individual matches will drift in real-time with the score—that will be the most vivid live case study of "how prices form."
Act IV · Variables Beyond Price: Regulation
Price is also influenced by a non-market variable: regulatory uncertainty.
On May 18, 2026, the Governor of Minnesota signed the SF4760 bill, making it the first US state to classify operating and advertising prediction markets as a felony (effective August 1, 2026). The CFTC (Commodity Futures Trading Commission) sued within 24 hours; Kalshi sued on May 28. CFTC Chairman Michael Selig's statement was:
"This Minnesota law turns lawful operators and participants in prediction markets into felons overnight."
Behind this lies an unresolved jurisdictional battle: the Third Circuit Court of Appeals ruled in favor of Kalshi on April 7 (event contracts are derivatives under CFTC jurisdiction), while the Ninth Circuit Court heard Nevada's appeal on April 16, leaning towards Nevada—the split between the two circuit courts may ultimately go to the Supreme Court. As of now, 17 states are challenging prediction market operators, 14 states have relevant legislation; Spain ordered ISPs to block Polymarket and Kalshi in 2026.
A strict distinction must be made here: Prediction markets follow the federal regulatory path of CFTC event contracts, while sports betting follows the state-licensed path—the same World Cup contract has drastically different legality across different jurisdictions. Regulatory uncertainty itself is a variable behind the price.
Epilogue · Returning to Those Two Numbers
Back to the opening—"France 17%" and "Spain 16.1%".
Now you know how these two numbers came about: one is the price cleared by hundreds of millions of dollars in trading volume on the market, subject to longshot bias and liquidity depth; the other is a frequency derived from simulating the entire World Cup ten thousand times by a supercomputer, subject to model lag and partially incorporating market information.
Which is more accurate? No rigorous cross-tournament comparison can answer that. After the World Cup concludes and contracts settle on July 20, we will publish a retrospective analysis—examining what the market and the model each got right and wrong.
Until then, whenever you see a "championship probability," it's worth asking: How was this number produced?
