Each published entry begins with one exact-score forecast. The site then derives logically consistent companion markets from that same scoreline, including match result, BTTS, Over/Under 2.5, FT Goals and Winning Margin.
Score probability model
Where sufficient recorded market prices exist, Correct-Score.net fits a Poisson goal model with a Dixon–Coles low-score adjustment to the recorded 1X2 and goal-market probabilities.
The output includes expected goals and a ranked score distribution. These are model estimates, not empirically calibrated probabilities, and they do not replace the published prediction or guarantee an outcome.
The model is not shown when the required market inputs are incomplete. Correct-score odds themselves are not used to force the published score to rank first.
Fair odds and model-market gap
When a published score has both a CSM-1.0 estimate and a recorded correct-score price, Correct-Score.net can convert the model estimate into model fair odds using 1 ÷ estimated probability and compare it with the recorded bookmaker price.
The bookmaker implied probability shown for a correct-score price is the raw reciprocal 1 ÷ decimal odds. Because a full correct-score market book is not always stored, no claim is made that this single implied probability is overround-free.
Until CSM-1.0 is empirically calibrated, the difference is labelled model-market gap rather than expected value. Historical ROI by gap band is diagnostic research, not proof of future profitability.
Starting point
Each published match begins with an exact-score forecast. That score is the central prediction; companion markets are used to explain what the score logically implies.
How companion markets are derived
A 2-1 forecast implies a home win, BTTS Yes, Over 2.5 goals, three total goals and a home winning margin of one goal. A 0-0 forecast implies a draw, BTTS No, Under 2.5 goals and zero total goals. This keeps the markets internally consistent.
What confidence means
Confidence is a relative editorial or model label used to distinguish selections on the same card. It is not a calibrated probability, a promise of success or a recommendation to increase stake size.
Inputs and limitations
Fixture information, available odds and structured match data can inform the workflow. Football remains affected by late injuries, rotations, weather, tactical changes, red cards and other events that cannot be predicted reliably.
When information is missing
We prefer a clearly derived market or an explicit “not available” state to invented data. Match-analysis text should not claim specific injuries, line-ups or recent form unless those facts are actually available to the publishing workflow.
Using the prediction
Readers can compare the exact-score view with the related markets and current bookmaker prices. A forecast is a starting point for analysis, not a guaranteed outcome.