Betting on the Draw in Soccer: The Group-Stage Edge
Betting on the draw soccer markets: why favorite-longshot bias misprices the draw, when biscotto group-stage incentives make it likely, and how to price it fair.
The draw is the only outcome in a soccer match that nobody walks in wanting to back, and that is precisely why it is the most systematically mispriced leg of the 1X2 market. Casual money flows to teams — fans pick a side, pick a winner, and pick a story. The draw has no fan base, so its price gets left to drift, and betting on the draw soccer markets is where the structural slack in 1X2 pricing collects.
Here is the mechanism in one sentence: the favorite-longshot bias pushes too much money onto both teams, which inflates the home and away prices and quietly leaves the draw too cheap. Across competitive group-stage fixtures, the draw lands around 26–30% of the time, yet the de-vigged market price routinely sits a point or two below that. With WC26 kicking off June 11 and 72 group-stage matches on the board, the draw is the single most repeatable value contract in the tournament — and the format gives it an extra push you can game.
Why the draw is the most mispriced 1X2 outcome
In a three-way 1X2 market, the home win, draw, and away win are mutually exclusive and their fair probabilities sum to 100%. Bookmakers add a margin (the overround) so the three prices sum to 104–108%, and that margin is not spread evenly. It is loaded disproportionately onto the outcomes the public wants to buy — the two teams.
Favorite-longshot bias, in money flow
Recreational bettors back winners. They take the favorite to assert itself or the underdog to spring a story, and almost nobody walks up to a window asking for the draw. That lopsided flow lets the book shade the home and away prices outward and keep the draw price soft, because soft is where the unbet outcome naturally settles.
The result: when you strip the vig, the draw is the leg most often left above its market-implied value. You are not finding a mispriced team and fading the crowd; you are buying the outcome the crowd structurally ignores.
De-vig the three-way to find the real draw price
You cannot judge a draw price by eye. Take an illustrative competitive group match: a moderate favorite at home against a well-organized opponent.
De-vig the 1X2 market to price the draw
Multiplicative devig. The fair column is what your model has to beat — not the raw price.
The three prices sum to 50 + 30 + 28 = 108¢, an 8% overround. De-vigging normalizes them back to 100%, so the vig-free probabilities are roughly Home 46.3%, Draw 27.8%, Away 25.9%. The screen sold you the draw at 30¢, but its fair value after stripping margin is only about 27.8% — at first glance the draw looks expensive, not cheap.
That is the trap, and the lesson: the raw 30¢ tag is mostly vig, not value. The real question is whether your model's draw probability beats the de-vigged 27.8%, because that — not the sticker price — is what you must outpredict.
“The draw's price is mostly margin. Strip the vig, then ask whether your model clears the number underneath.”
Where your model beats the de-vigged number
Two conditions reliably push the true draw probability above that de-vigged figure. First, evenly matched sides: the closer the two teams in strength, the higher the draw rate — a coin-flip matchup peaks the draw near 28–30%. Second, low expected goals: tight, low-event games cluster outcomes around 0-0 and 1-1. When you expect a defensive group match — the same low-total read behind the Over/Under and BTTS markets — the draw is mechanically more likely than a goal-fest.
So the draw edge and the Under edge are the same family of bet: both pay off when the match has fewer goals and less separation than the public expects.
When group-stage incentives make a draw likely
The group stage hands you something the knockouts never can: matches where a draw is the mutually preferred result. This is the famous biscotto — Italian betting slang for a "cookie," a quietly convenient drawn scoreline that sends both teams through.
The mutual-qualification scenario
Picture the final round of a group. Two sides walk into their head-to-head knowing that a draw qualifies both of them, while a winner-take-all result would eliminate one. The third team in the group can only catch them by winning big elsewhere. The incentive structure is brutally simple: both teams have everything to gain from not pushing for a winner.
You do not need to allege a fix to price this. The motivation is real and legal — a side resting starters and protecting a point it is happy to take is just rational. When the table sets up a mutual-qualification draw, the true draw probability can climb well above 30%, sometimes toward 40%, while the market — pricing the game as a normal contest — lags.
Dead rubbers cut the other way
Be disciplined: not every late group game favors the draw. A dead rubber where one side has clinched and rotates heavily can become more decisive, not less, if the motivated opponent runs over a second-string lineup. The biscotto raises draw probability; an asymmetric dead rubber can lower it. Read the incentives of both teams before you assume a draw — the format manufactures both setups, and the related to-advance market breakdown walks through how clinching scrambles motivation.
A worked example: is the draw +EV?
Put it together. Suppose you have a tight, low-scoring group match between two evenly matched sides, and the table gives both a comfortable point. Your model lands the draw at a fair 30%. The de-vigged market — from that 1X2 board above — has it at about 27.8%, and you can buy the draw contract at 27¢.
Is this contract +EV?
EV is only as good as your probability. Garbage-in, garbage-out — devig the market and pressure-test your model.
The math: EV per contract = 0.30 × (1 − 0.27) − 0.70 × 0.27 = 0.219 − 0.189 = +$0.03 per contract, an +11% return on risk. The edge is the gap between your 30% read and the 27¢ price — modest per contract, but it is a repeatable edge you can take across dozens of group fixtures, which is exactly what a structural bias is worth.
The draw: market price vs model
The biscotto row is where the edge gets fat: a 31¢ price against a 39% model is an 8-point gap, far beyond the routine 2–3 points. That is the difference between grinding a small structural lean and pressing a genuine mispricing the market hasn't caught up to.
How to actually trade the draw at WC26
A short, repeatable playbook:
- De-vig every 1X2 before judging the draw. The raw 30¢ tag is mostly margin. Strip it to the ~27–28% underneath, then compare to your model.
- Buy the draw in even, low-goal matches. Closely matched sides plus a suppressed total is the bread-and-butter setup. The draw and the Under are the same low-event thesis.
- Map Matchday 3 for biscotto. Any fixture where a draw qualifies both teams is your highest-conviction draw of the group stage. Price those before the market wakes up.
- Respect the dead-rubber exception. A clinched side rotating against a motivated opponent can make the game more decisive. Read both teams' incentives.
- Cross-check the moneyline. The draw price has to be consistent with the home and away legs — when the 3-way moneyline moves on team news, the slow-to-adjust draw leg is often the trade.
Nobody walks into the World Cup wanting to back a 1-1. That is the entire opportunity. The crowd will keep buying teams and stories, and the draw will keep settling a couple of points light — and on Matchday 3, when the table quietly turns a head-to-head into a shared point, the cookie is yours for the taking.
Frequently asked
Why is the draw underpriced in soccer betting?
What is a biscotto in World Cup betting?
How do you price a fair draw probability?
Is betting on the draw actually profitable?
When is the draw most likely in the group stage?
How do I de-vig a draw price?
Sources (5)
- Polymarket — 2026 FIFA World Cup Winneraccessed 2026-06-06
- Kalshi — Sports event contractsaccessed 2026-06-06
- FIFA — 2026 World Cup formataccessed 2026-06-06
- Pinnacle — World Cup 1X2 marketsaccessed 2026-06-06
- FBref — Squad and xG dataaccessed 2026-06-06