Correct Score Trading: How the Market Prices a Scoreline

A correct score market looks like a long list of prices. It is really one probability distribution, chopped into pieces. Understanding how those pieces are cut, and how they move together, is the whole of correct score trading. Everything else is execution.

One thing to state at the outset: our own dataset does not record correct score markets. Nothing in this piece is a measurement of what happens in practice, and no claim here is backed by our numbers. It is a description of mechanics, with worked arithmetic from a standard scoring model. Real markets deviate from that model, sometimes a lot.

What the market actually contains

On Betfair Exchange, a football correct score market typically lists every scoreline up to three goals per side — 0-0, 1-0, 2-0, 3-0, 1-1, 2-1, 3-1, 2-2, 3-2, 3-3 and the away-side mirrors — plus three catch-all runners: any other home win, any other away win, any other draw. Roughly twenty runners in total.

The buckets exist because the tail is infinite. There is no runner for 5-2, so 5-2 lives inside “any other home win”. Those buckets are the hardest runners to price precisely, because each one aggregates dozens of unlikely results into a single number.

Each runner can be backed or laid, like any exchange market. Laying 1-1 is a bet that the final score is anything other than 1-1. That sounds generous until the price is examined.

Why the prices have to add up

The scorelines are mutually exclusive and, with the buckets included, exhaustive. Exactly one of them settles as a winner. So their implied probabilities must sum to 100% — plus the overround, the excess that makes the book profitable for the layers in aggregate.

That constraint ties the correct score market to the match odds market. Sum the implied probabilities of every home-win scoreline plus “any other home win”, and the result should match the home win probability in the match odds market. If it does not, the two markets disagree, and one of them is mispriced. On liquid fixtures, arbitrage between them keeps the gap small and short-lived.

The same logic binds correct score to over/under 2.5 goals. 0-0, 1-0, 0-1, 1-1, 2-0 and 0-2 together are the under 2.5 market. Anyone forming a view on a scoreline is implicitly forming a view on totals and on the result at the same time.

Where the low scorelines come from

The standard starting point is a goal expectancy for each side, run through a Poisson distribution. Take a fixture with home expectancy 1.5 goals and away expectancy 1.1. Treat the two as independent — a simplification, and a known weak point, but it gets the shape right.

Probability of the home side scoring exactly 0, 1, 2, 3 goals: 22.3%, 33.5%, 25.1%, 12.6%. Away side: 33.3%, 36.6%, 20.1%, 7.4%.

Multiply the pairs:

Scoreline	Probability	Fair odds
0-0	7.4%	13.5
1-0	11.1%	8.98
0-1	8.2%	12.2
1-1	12.3%	8.16
2-0	8.4%	11.96
2-1	9.2%	10.88
0-2	4.5%	22.2
1-2	6.7%	14.9

Aggregate the lot and the model gives home win 46.4%, draw 25.8%, away win 27.8% — a match odds line of about 2.16 / 3.88 / 3.60 before overround. The home scorelines the market lists (1-0 through 3-2) account for 40.0 percentage points of that 46.4%, so “any other home win” carries the remaining 6.4%, or around 15.6. “Any other draw” is worth about 0.1% and is priced in the hundreds.

Notice how flat the top of the distribution is. The four most likely scorelines sit between 8% and 13%. No single scoreline is ever a favourite in any meaningful sense. That flatness is why correct score prices are long, why the spreads matter so much, and why a small error in goal expectancy shifts the ranking of the runners entirely.

One goal, and the whole book reprices

This is the mechanism every correct score approach depends on, so it is worth doing in numbers.

Same fixture. Twenty minutes gone, no goals. Seventy minutes remain, so scale the goal expectancies down in proportion to time left: 1.167 for the home side, 0.856 for the away side. The comparison that matters is not with the pre-match prices but with the prices at that moment, goalless, an instant before anything happens:

Scoreline	20’, 0-0	20’, after a home goal
0-0	7.56	dead
1-0	6.48	7.56
2-0	11.10	6.48
2-1	12.98	7.57
1-1	7.57	8.83
0-1	8.83	dead

Now the home side scores. Every scoreline with a home total of zero is dead: 0-0, 0-1, 0-2, 0-3 and part of the away bucket — together 31.1% of the book’s probability a moment earlier. On the exchange those runners do not vanish; their back prices run to the maximum of 1000 and the lay side empties.

The survivors absorb that 31%, and not in the way intuition suggests:

• 1-0 drifts, from 6.48 to 7.56. Before the goal it needed exactly one home goal and none from the away side. After it, it needs no more goals from either side — and a home team expected to score 1.17 more goals is less likely to stop than it was to score exactly once. The goal that created the scoreline made it less likely to be the final one.
• 2-0 nearly halves, from 11.10 to 6.48, because it now needs exactly one more home goal instead of two.
• 2-1 shortens from 12.98 to 7.57 for the same reason.
• 1-1 drifts from 7.57 to 8.83: it now needs an away goal and no further home goal.
• The draw as a whole falls from about 30% to 18%. The home win rises from 43% to 73%.

The lesson is that a goal is not a smooth move. It is a discontinuity. Almost a third of the book is extinguished in one tick and redistributed across the survivors in unequal proportions. Some runners shorten sharply, some drift, and the scoreline that has just appeared on the board is not necessarily among the winners. A position in this market is a position on which side of that redistribution it lands.

How traders use the market

Laying a scoreline

Laying 1-1 at 8.16 wins in about 88% of cases. It also carries 7.16 units of liability for every unit of return. The payout profile is many small wins and occasional large losses, and at a fair price the two exactly offset. There is no edge in the shape of the bet itself; the only possible edge is a better estimate of the true probability than the price implies.

What it needs: a genuine reason to think the scoreline is less likely than priced — a team that rarely draws, a fixture with an unusual goal profile, a red card. What breaks it: the loss arrives on the single most likely outcome in the book, and at 8.16 one losing result wipes out the gains of about seven winning ones.

Backing a cluster of scorelines

A trader expecting a comfortable home win might back 1-0, 2-0 and 2-1 together. Combined fair probability 28.7%, which is an effective price of 3.49 with equal-return staking. In a real market with overround, those three runners might be available at 8.6, 11.5 and 10.4, giving an effective 3.34.

Compare that to backing the home win in match odds at around 2.10. The cluster pays more because it is a narrower claim. The trader is buying specificity, and paying the overround three times to do it. It wins only if the exact scoreline lands inside the chosen set — a 2-2 or a 3-1 loses the entire stake even if the reading of the match was broadly correct.

Trading the decay of 0-0

While a match stays goalless, 0-0 shortens continuously. On the model above, 0-0 opens at 13.5, is worth about 3.67 at half-time if the score is still level, and about 1.78 with twenty minutes left. Backing early and laying back later locks a position across that decay.

The profile is brutally asymmetric. The gain accrues slowly, minute by minute, and most of it arrives in the last half hour. The loss arrives instantly and in full, at any moment, and a goal in the 85th minute destroys a position that was worth nearly all of its potential value a second earlier. There is no partial exit once the ball crosses the line.

Two further problems. First, naive time-scaling understates late goals — scoring rates rise in the second half — so a model that divides expectancy proportionally by minutes will systematically overvalue a late 0-0. Second, anyone following the match on a delayed picture is trading against people who see it sooner. An offer left in the market is most likely to be taken by someone who already knows it is a bad price for the person who left it.

Liquidity is the binding constraint

Outside the largest fixtures, correct score markets are far thinner than match odds. The money in a match is spread across twenty runners instead of three, and most of it goes to the result market anyway.

Then there are tick sizes. Between 6 and 10, prices move in steps of 0.2. Between 10 and 20, steps of 0.5. Between 20 and 30, steps of 1. A runner quoted 8.2 back, 8.6 lay has a round-trip cost of roughly 5% before anything happens in the match. At 20.0, the next tick up is 21 — a single tick is 5%. Commission on net market winnings sits on top.

An edge of two or three percent in probability terms, which is already an ambitious claim for a manual trader, does not survive a 5% spread. This is the point at which most correct score ideas fail: not in the reasoning, but in the fill.

Liquidity also collapses precisely when it is most needed. After a goal, the market is suspended, reformed and repriced, and the first prices back are wide. A trader wanting out of a damaged position at that moment will pay for the privilege.

What the market is good for

Correct score is a precise instrument. It expresses views that match odds and totals cannot: a specific margin, a specific scoreline pattern, a specific way a match ends. That precision is genuinely useful for anyone with a detailed model.

It is also expensive, thin and discontinuous. Positions do not unwind gracefully. Prices move in jumps. Spreads are wider than in the headline markets, and at long prices a single tick is a large share of the stake. None of that makes the market unusable, but it does mean the arithmetic has to be done before the position is taken, not after.

---

This content is provided for informational purposes only. It does not constitute betting advice or a recommendation to gamble. Gambling can be addictive; play responsibly. 18+.