Onchain Sports Betting: How to Bet the FIFA World Cup Self-Custodially

The 2026 FIFA World Cup didn't just break attendance records — it broke onchain betting records. On opening day Polymarket processed $118M in volume, and its "World Cup Winner" market has crossed roughly $2 billion in cumulative trading. FIFA itself appointed an official prediction-market partner. If you've wondered how people bet on sports without handing a bookmaker their balance — and what the catch is — this is the deep version: how the plumbing actually works, the three sportsbook architectures, the oracle risk almost nobody explains, the vig math, and the legal reality.

What "onchain" actually changes

A traditional sportsbook is a trusted intermediary: it holds your deposit, sets the odds, decides who won, and you trust it to pay out and to let you withdraw. Every one of those points is a counterparty risk — frozen accounts, limited winners, insolvency (ask anyone who had funds on a collapsed CeFi platform). Onchain betting re-architects the stack:

• Self-custody. You bet from your own wallet. The protocol escrows the stake in a smart contract; it never sits on a company balance sheet. It can't quietly limit you for winning or vanish with your funds.
• Lower vig (the bookmaker's margin). Legacy books bake in a 4.8–7% hold. A peer-to-peer onchain exchange like SX Bet charges 0% on single bets, because you're betting against other users, not the house. More of every dollar stays with bettors.
• Transparency and verifiability. Odds, liquidity depth, every trade, and the final settlement are on-chain. You can audit the order book and the resolution instead of trusting a screen.
• Composability. Positions are tokens. You can sometimes sell a bet back before the event ends, hedge it elsewhere, or provide liquidity to be the house.

The trade-offs are equally real, and we'll be blunt about them: more setup friction, smart-contract and oracle risk, thin liquidity outside the flagship markets, and genuine regulatory uncertainty. Onchain removes the bookmaker's discretion; it does not remove risk.

The two models (and why it matters)

Almost everything in this space is one of two things. Confusing them is the most common beginner mistake.

Model 1 — Prediction markets (Polymarket)

A prediction market turns each outcome into a tradeable yes/no share that settles at $1 if it happens and $0 if it doesn't. If "Argentina to win the World Cup" trades at 28¢, the market is pricing a ~28% probability. Buy it; if Argentina wins, each share redeems for $1. Crucially, you can sell before the event resolves — if Argentina's price climbs to 45¢ after a strong group stage, you can take profit without waiting for the final. It's a market, so you're trading a probability, not just placing a static bet.

Under the hood, Polymarket runs a central limit order book (CLOB): you trade peer-to-peer against other users' bids and asks — not against the house — with a transparent book and live price discovery. Order matching happens off-chain for speed, but settlement is on Polygon in USDC, and the outcome shares are minted via the Gnosis Conditional Token Framework. The scale is now enormous: ~2.4 million traders and ~$62B of cumulative volume across sports, elections, crypto, and culture.

Model 2 — Onchain sportsbooks (three different engines)

These feel like a traditional book — moneylines, spreads, totals, parlays — but settle on-chain. The important nuance is that "onchain sportsbook" hides three completely different architectures, and the design determines your odds, fees, and risk:

a) Peer-to-pool — Azuro (the vAMM "house" model). Azuro is infrastructure, not a consumer app: a single concentrated liquidity pool + a virtual AMM and a "LiquidityTree" that powers dozens of betting front-ends. Odds are computed dynamically from the pool's funds and "virtual reinforcements" (liquidity virtually assigned to each outcome). Liquidity providers act as the house and earn the margin; bettors get deep, always-on markets without an order book. Its AZUR token (launched June 2024) governs the protocol via the Azuro DAO.

b) AMM — Overtime (built on Thales). Overtime uses a Sports AMM and a Parlay AMM: odds move based on the balance of the liquidity pool rather than a book, so there's always a position available to take, and LPs underwrite the action. It's crypto-native, peer-to-AMM, has done $200M+ in volume, and generally offers better odds than legacy books. Governance/incentives run through the Thales/Overtime token (OVER).

c) Peer-to-peer exchange — SX Bet (the betting exchange). SX Bet is a true exchange order book: one bettor posts an order at the odds they want, another takes the opposite side, and the protocol's matching engine escrows both stakes. Because you're betting against other users, not the house, the fee is 0% on single bets (5% on parlay profit, which uses a different mechanism) — an effective vig under 0.5% versus the ~4.8% industry norm, roughly a 90% cut. It runs on SX Network (its own chain, migrating to an Arbitrum rollup); the SX token pays gas, bonds validators, and powers governance.

At a glance — which is which

The part nobody explains: how outcomes resolve

Here's the question that actually matters and that most guides skip: after the match, who decides who won, and what if they're wrong? With a traditional book, the company decides. Onchain, an oracle does — and this is where the subtle risk lives.

Polymarket uses the UMA Optimistic Oracle. The flow: once an event ends, anyone can propose the outcome by posting a bond; a ~2-hour challenge window opens; if nobody disputes, the market resolves and winning shares redeem at $1. If someone disputes, it escalates to UMA's Data Verification Mechanism (DVM) — a vote of UMA token holders, weighted by holdings. "Optimistic" because the proposal is assumed correct unless challenged.

For a clean question — "did Brazil beat Croatia?" — this is robust. The danger appears with ambiguous wording and concentrated voting power:

• Governance attacks are real. In March 2025, a single actor used ~25% of UMA voting power to force a $7M market — "Will Ukraine agree to Trump's mineral deal before April?" — to resolve "Yes" despite no agreement; the odds were pushed from 9% to 100%. Polymarket called it "unprecedented." The structural problem: roughly four wallets control ~40% of UMA.
• It's getting more common. Polymarket has seen 1,150+ disputed markets in 2026 alone, already past the full-year 2025 total. The May 2026 MicroStrategy "did it sell BTC?" dispute drew $60M+ in volume — the largest since the Ukraine market.
• Resolution-rule ambiguity. Traders have accused Polymarket of retroactively adjusting timing language after markets closed. The lesson: the exact resolution criteria are the product. Read them before you bet.

Onchain sportsbooks lean on different oracles — Chainlink sports data feeds are common (and are what FIFA's official partner ADI PredictStreet uses) — which are less prone to governance games for clean game results, but introduce their own data-source trust. Takeaway: on any platform, the question "how does this exact market resolve?" matters more than the odds.

How to bet onchain, step by step

Step 1 — Confirm it's legal where you are

Non-negotiable first step. Online betting and prediction markets are restricted or illegal in many jurisdictions, platforms geoblock regions, and the rules change. Onchain access does not make it legal where it isn't.

Step 2 — Set up a self-custody wallet

You bet from your own wallet, so start there. MetaMask is the default across Polygon, Arbitrum, Optimism, Base, and the other chains these venues use.

Step 3 — Get USDC at the best price

Onchain betting settles in USDC. If you're holding another asset, don't overpay converting it — CoW Swap batches your order, makes solvers compete, protects you from MEV, and often beats a direct AMM price (you can even pay gas in the token you're selling).

Step 4 — Bridge to the venue's chain

Match the chain to the venue: Polygon for Polymarket, Arbitrum for SX/Overtime, or Base/Optimism/Gnosis for various Azuro-powered books. Move USDC cheaply with an intent bridge.

Step 5 — Pick your model and place a position

Connect your wallet and bet the way that fits you. Want to trade probabilities and exit early? Use Polymarket (and read our Polymarket guide). Want traditional odds with a low vig? Use an Azuro-powered book, Overtime, or SX Bet. Start small on any new platform, and — say it again — read how the specific market resolves before you stake.

The vig and EV math (read this before you size up)

The seductive pitch is "lower vig = better odds." True, but understand what it does and doesn't buy you:

• Vig sets your break-even win rate. A standard −110 line (≈4.5% vig) needs you to win ~52.4% of bets just to break even. Cut the vig toward ~0%, and break-even drops toward ~50%. That's a real edge over a lifetime of bets — but it makes a coin-flip break-even, not profitable.
• The expected value of a typical bet is still negative. Onchain shrinks the house edge; it does not hand you predictive skill. Most bettors lose over time, lower vig or not.
• The genuine edges are structural, not magic. On an exchange (SX), sharp bettors can post their own odds and get matched rather than taking the book's price. On prediction markets, you can trade out of a position or arbitrage price gaps between Polymarket and a sportsbook. These are skill-and-effort edges, not free money.

Bottom line: budget it as entertainment you expect to lose, and treat any structural edge as a way to lose slower, not a salary.

The 2026 World Cup, onchain

The tournament was the genre's coming-out party. Polymarket ran 380+ live World Cup markets, did $118M on opening day, and its flagship "World Cup Winner" market reached roughly $2B cumulative — at one point pricing the field so tightly that "the favorite" was effectively a coin toss. Most tellingly, FIFA appointed ADI PredictStreet as an official prediction-market partner, using Chainlink oracles to verify outcomes — a governing body putting its name on onchain markets, which would have been unthinkable a cycle ago. (For where this fits among non-onchain venues like Kalshi, see our best prediction markets guide.)

The regulatory reality

This is the part that can actually cost you, independent of any bet:

• It's jurisdiction-by-jurisdiction. Sports betting and prediction markets are legal in some places, restricted in others, and outright illegal in many. US treatment varies by state and by product (sports vs event contracts).
• Polymarket's US history is instructive. It paid a CFTC settlement and blocked US users for years, only returning to the US in late 2025 as a CFTC-licensed exchange. Access today still depends on where you are; many platforms geoblock.
• Integrity probes are live. Regulators are scrutinizing insider-trading risks on prediction platforms (people betting on information they shouldn't have). Don't assume "decentralized" means "unregulated and consequence-free."
• Tax applies. In most jurisdictions, gambling winnings and/or crypto disposals are taxable events — and every on-chain bet and USDC swap may create a reportable transaction. See our crypto tax guide and keep records.

The risks, summarized

• It's gambling. Negative expected value for most bettors, lower vig or not.
• Resolution / oracle risk. Ambiguous wording + concentrated oracle voting power can mis-settle a market. Read resolution rules; prefer clean, liquid markets.
• Smart-contract risk. Funds sit in contracts; use established, audited protocols.
• Liquidity risk. Thin markets mean bad fills and slippage; flagship World Cup books are the exception, not the rule.
• Legal / tax risk. Restricted in many places; winnings and swaps may be taxable.

The workflow, at a glance

1. Check it's legal where you live.
2. Wallet → MetaMask (self-custody).
3. Get USDC → CoW Swap for the best price.
4. Bridge → deBridge to Polygon / Arbitrum.
5. Bet → Polymarket (prediction market) or Azuro / Overtime / SX Bet (sportsbook) — after reading the resolution rules.

For more, see our Polymarket guide, best prediction markets, best crypto bridges, and crypto tax guide.

Not financial, legal, or betting advice. Gambling carries a real risk of loss — if it stops being fun, stop. Confidential help is available through your local responsible-gambling service (e.g. the National Council on Problem Gambling, 1-800-GAMBLER in the US).