Correct Score

Correct score is the market people bet when they want the payout to be interesting. It is also the market where the price you are offered is furthest from the price you should be getting, and those two facts are related.

How the market settles#

Straightforward, with a couple of edges worth knowing.

The bet settles on the score after 90 minutes plus stoppage time. Extra time and penalties do not count, so a tie that finishes 1-1 and goes to a shoot-out settles as 1-1 no matter who lifts the trophy. Own goals are credited to the team that benefits, exactly as on the scoresheet. And direction matters: 2-1 to the home side and 2-1 to the away side are two different bets at two different prices.

Books list perhaps twenty to thirty scorelines individually and then bundle the rest into "any other score" or "any other home win". If your bet is 6-4 you are buying that bundle, not a specific result.

Why the likeliest scoreline is still a long shot#

Take an ordinary fixture. The home side is expected to score 1.50 goals, the away side 1.10. Treat the two as independent Poisson processes — the standard starting point for goal modelling — and every scoreline gets a probability.

Read the top row again. The single most likely outcome of the match — the one you would name if forced to pick — happens about 12% of the time. Fairly priced, that is 8.16, because 1 ÷ 0.123 = 8.16. Eighty-eight times in a hundred, the best possible guess is wrong.

That is not a defect in the model. It is the market. Football produces a genuinely wide distribution of scorelines, and no amount of insight collapses it into a confident single answer. Anyone selling you a correct-score tip with conviction is selling conviction, because the underlying probability does not support it.

Move the sliders in the Poisson calculator below to your own fixture and read the "most likely scorelines" strip underneath the grid. To reproduce the table above exactly, set 1.50 and 1.10 and untick the Dixon-Coles box; then tick it again and watch the four low-score cells shift — that correction exists precisely because raw Poisson gets 0-0, 1-0, 0-1 and 1-1 slightly wrong.

Push the goal rates up and something instructive happens: the peak gets lower. A high-scoring fixture spreads its probability across more plausible scorelines, so the best available guess becomes worse, not better. The matches that feel most predictable in the 1X2 market are often the least predictable here.

Every cell of the grid is a correct-score price#

This is the part worth internalising, because it reframes the whole market.

A goal-based model does not price 1X2 and then price correct score separately. It builds one grid of scoreline probabilities and reads every market off it. Home win is the sum of the cells below the diagonal. Over 2.5 is the sum of every cell with three or more goals. Both teams to score is everything outside the top row and left column. And correct score is simply the grid itself, uncollapsed — one cell, one price.

Which means correct score is the most demanding test of a goal model there is. The 1X2 numbers survive quite a lot of error, because summing dozens of cells lets individual mistakes cancel out. A single cell has nothing to hide behind. Our models produce that whole grid for every fixture, and we publish accuracy on the aggregate markets rather than the cells, for exactly this reason.

The margin is the real problem#

The prices look generous. They usually aren't, and you can check that yourself in about a minute.

Take a book's full correct-score list, convert every price to an implied probability with 1 ÷ decimal odds, and add them up. On a 1X2 market that total normally lands a few percent above 100%. On a correct-score list it tends to be far higher — and the whole of that excess is the operator's cut.

There is a reason for the width. A book quoting thirty separate lines on a low-liquidity market has thirty separate ways to be wrong, and it charges for that risk. But knowing the reason does not make the toll smaller. If you want to strip the margin from a book properly rather than by eye, the fair odds calculator does it with the power method, the same one Tofiko applies to every market price it displays.

Where it might be worth a look#

Not as a staple. The margin makes correct score a poor place to grind, and the variance means you could be right about the process for a very long time before the results agree — which is the sample-size problem in its most brutal form, since you are now waiting for outcomes that only arrive one time in ten.

Where it becomes interesting is at the extremes of the grid. Books often price the headline scorelines carefully and the unusual ones by rule of thumb. A fixture with a genuinely lopsided goal expectation — a heavy favourite against a side that concedes freely — pushes probability into cells the market may still be pricing off a generic template. Whether that gap is real is an empirical question, and it is exactly the sort of thing that would show up in closing line value long before it showed up in profit.