Kelly Criterion Sports Betting: World Cup 2026 Bet Sizing

Two traders find the exact same edge: Brazil at 13¢ on a true probability of 15%. One bets 40% of their bankroll because they're "confident." The other bets 4%. Thirty-nine days and one group-stage upset later, the first trader is busted and the second is still trading. Same edge, opposite outcomes — because sizing, not selection, is what kills bankrolls at a World Cup.

The Kelly criterion is the math that tells you how much to stake once you've found an edge. It's the bridge between "this contract is +EV" and "this is exactly how many dollars I can put behind it without blowing up." For 2026 — 48 teams, a sprawling bracket, and a month of correlated positions repricing daily from the June 11 opener to the July 19 final — getting the size right matters as much as getting the read right.

The Kelly criterion for binary contracts

Kelly answers one question: what fraction of your bankroll maximises your long-run growth rate? Bet too little and your edge never compounds. Bet too much and variance grinds you toward zero — even with a real edge.

For a binary Kalshi or Polymarket contract, the formula is clean. A contract bought at price c cents pays $1, so the net decimal odds are b = (100 − c) / c — your profit per dollar risked. With your fair probability p (and q = 1 − p):

Kelly fraction f* = (b × p − q) / b

That's identical to (p − c/100) / (1 − c/100) — your edge, scaled by how much you stand to win. The more your p exceeds the price, the larger the optimal stake. Note immediately: when p equals the price, f\* is zero. No edge, no bet. Kelly never tells you to stake on a fair-priced contract.

Working the Brazil contract through Kelly

Take that Brazil edge: price 13¢, fair probability 15%, bankroll $1,000. The decimal payout is b = 87/13 = 6.69. Then f\* = (6.69 × 0.15 − 0.85) / 6.69 = (1.004 − 0.85) / 6.69 ≈ 0.023.

Full Kelly says stake 2.3% of bankroll — about $23. Plug your own numbers into the calculator and watch how violently the recommended size swings with small changes in your fair probability.

Now do the experiment that every trader should do once: nudge the fair probability from 15% to 17%. Full Kelly roughly doubles. Drop it to 14% and the stake collapses toward zero. Kelly is extraordinarily sensitive to your probability estimate — and your estimate is the thing you're least sure about. That sensitivity is the entire argument for not betting full Kelly.

Why full Kelly is too aggressive

Full Kelly is growth-optimal only if your probability is exactly right and your bets are independent. Neither is true at a World Cup. Three problems compound:

• Your p is an estimate, not a fact. Kelly assumes you know the true probability. You don't — you have a model with error bars. If you overestimate your edge by even a couple of points, full Kelly overbets, and overbetting is far more punishing than underbetting.
• The variance is savage. Even with a correct edge, full Kelly produces drawdowns that are psychologically and financially brutal. A 50% drawdown is a coin-flip occurrence under full Kelly — and recovering from −50% requires a +100% gain.
• Overbetting past the optimum loses money. The growth curve is asymmetric. Bet a little under Kelly and you sacrifice a sliver of growth. Bet a little over and growth falls off a cliff; bet at twice Kelly and your long-run growth rate is zero despite a real edge.

Half-Kelly and quarter-Kelly: the survivable defaults

The fix is fractional Kelly — betting a fixed fraction of the full-Kelly number. The trade-off is remarkably favourable:

• Half Kelly captures about 75% of the growth rate of full Kelly while roughly halving the volatility. For most serious traders this is the default. Our Brazil bet drops from $23 to about $12.
• Quarter Kelly is the choice when you genuinely distrust your edges — early in the tournament, on thin markets, or when your model is unproven. You give up more growth but your drawdowns become tolerable.

Why half Kelly captures so much growth for so little risk is the key insight: the growth function is roughly parabolic near the optimum. Pull back to half the stake and you're still high on the curve — you've barely moved down in growth, but you've slid well down the risk axis.

Read the chart carefully: betting 1.5× Kelly delivers the same growth as half Kelly — but with vastly more variance. And 2× Kelly zeroes out your growth entirely. There is never a reason to bet above full Kelly, and excellent reasons to bet well below it.

Correlated positions across a 48-team bracket

Here's where naive Kelly quietly destroys World Cup traders. The formula assumes each bet is independent. Across a single bracket, almost nothing is.

If you hold YES on France to win the trophy, YES on France to win their group, and YES on a France striker for the Golden Boot, you don't have three separate Kelly bets — you have one giant correlated position on France. Size each at half Kelly independently and your true exposure is closer to full Kelly or worse. One bad France result detonates all three at once.

How to size correlated bets

• Treat correlated contracts as a cluster and apply your Kelly fraction to the cluster's total exposure, not to each leg. If France-related positions should be 5% of bankroll combined, that's the cap — split across the legs, not 5% each.
• Cap exposure per team and per match-day. A practical rule: no single team's cluster exceeds 10–15% of bankroll, no matter how many +EV France contracts you find. The group draw tells you which of your positions share a failure path.
• Watch negatively correlated hedges. Some positions partially offset — holding YES on two teams in the same group that can't both top it. These reduce variance and let you size up slightly, the mirror image of the correlation problem.

Surviving 39 days without going broke

The tournament is a marathon of sequential bets, and risk of ruin is the probability your bankroll hits zero before it can compound. Three habits keep you in the game from Mexico City to MetLife:

• Recompute Kelly off your current bankroll, every time. As the bankroll shrinks, your dollar stakes shrink with it — that's the built-in circuit breaker that makes Kelly mathematically impossible to bust from a single bet. Betting fixed dollar amounts throws that protection away.
• Floor your stakes. Even strong +EV spots get a hard cap — never let a single contract exceed, say, 5% of bankroll regardless of what Kelly says, because your edge estimate could simply be wrong.
• Stay liquid for the knockouts. The juiciest mispricings often arrive after the group stage, when upsets scramble every path. Don't deploy your whole bankroll in week one. Track the schedule and keep dry powder.

This sizing discipline only works on top of genuine edges — Kelly multiplies an edge, it doesn't create one. Pair it with the expected value framework for finding +EV contracts, go deeper on staying solvent in the bankroll management and risk-of-ruin guide, and shore up the fundamentals with the implied probability guide.

The trader who wins the World Cup isn't the one who calls the most upsets. It's the one who's still got chips on the table when the final whistle blows in July.