Football Score Probability Calculator
Enter pre-match football (soccer) odds to calculate market-implied correct-score probabilities, Home/Draw/Away win probabilities, BTTS and totals. Start with 1X2 prices; add Over/Under 2.5 odds when available.
Correct score calculator: from odds to score probabilities
Enter decimal odds, and the tool removes the market overround with proportional normalization, fits home and away goal rates (λ), and generates a full score probability matrix using Poisson or Dixon-Coles. It also derives Home/Draw/Away, BTTS, totals and model-implied fair odds from the same matrix.
Football win probability: sum the score cells where the home team scores more goals for Home, the diagonal for Draw, and the cells where the away team scores more for Away. These are pre-match model probabilities conditional on the supplied odds; fitted values can differ from the de-vigged market targets.
For a step-by-step explanation of the inputs and output, read how the correct score calculator works.
Need to convert a single betting price first? Use the Implied Probability Calculator for decimal, American or fractional odds, or to convert probability back to equivalent odds.
Input
Complete pre-match 1X2 decimal odds (required) + Over/Under 2.5 odds (optional)
Engine
Poisson (independent goals) or Dixon-Coles (low-score correction via ρ parameter)
Output
Score heatmap, Home/Draw/Away probabilities, top scorelines, BTTS, O/U 1.5–4.5, fair odds and model-fit residuals
This is not a prediction tool. It translates what the betting market already implies into a visual probability map.
How to use it
Step 1
Enter Home / Draw / Away decimal odds from one bookmaker at the same pre-match snapshot, with matching settlement rules.
Step 2
Optionally add both Over and Under 2.5 prices from the same snapshot and settlement period to help anchor total goal expectation.
Step 3
Choose Poisson for the independent baseline or Dixon-Coles for the local four-cell low-score correction. A non-zero ρ changes 0:0, 1:0, 0:1 and 1:1 directly.
Step 4
Read the heatmap, Home/Draw/Away probabilities, top scores, derived markets and Model Fit. Residuals describe input-to-model mismatch; they are not predictive accuracy.
You can name teams, use presets, and share any calculation via a unique URL.
What the numbers mean
Heatmap — estimated probability of each scoreline. If 1:1 shows 12.2%, the market-implied chance is roughly one in eight.
Fair odds — model-implied break-even prices after the current proportional de-vig baseline. For complete-market margin removal and alternative methods, use the No-Vig Calculator.
BTTS — sum of all scorelines where both teams score; see the Both Teams to Score probability guide for the Poisson formula, Dixon-Coles adjustment and fair odds. Over/Under — sum of scorelines above the threshold. For one-team goal lines such as Team Over 0.5, 1.5 or 2.5, use the team-total probabilities and fair odds.
Model Fit — residuals between the de-vigged market targets and the model output. Smaller residuals mean a closer numerical fit to the supplied markets; they do not measure future predictive accuracy.
Worked example: market probabilities and an illustrative score matrix
Home
2.10
Draw
3.50
Away
3.40
Over 2.5
1.85
Under 2.5
2.00
The 1X2 booksum is 105.60%, an overround of 5.60 percentage points. Proportional margin removal gives Home 45.09%, Draw 27.06% and Away 27.85%. De-vigging the totals market separately gives Over 2.5 = 51.95% and Under 2.5 = 48.05%.
To illustrate the score-matrix calculation, the table below uses fixed Dixon-Coles parameters λhome=1.56, λaway=1.19 and ρ = −0.03. This parameter example is not an exact fit to all five prices above; the live fitted result should be read alongside its Model Fit residuals.
| Score | Prob |
|---|---|
| 1 : 1 | 12.2% |
| 1 : 0 | 9.6% |
| 2 : 1 | 9.3% |
| 2 : 0 | 7.8% |
| 0 : 1 | 7.3% |
In this example, the Top 5 carries only part of the probability mass; the rest is distributed across many other scorelines. Ranking a few scores does not turn them into high-probability predictions.
Why the most likely score is still unlikely
Reality check
Even the highest-probability exact score is usually only one small part of the full distribution. A score can rank first while most probability still lies on other outcomes. Historical Top-k coverage is examined separately in the Most Likely Score Accuracy study.
"This is the least unlikely outcome among many unlikely outcomes." Understanding that distinction is the entire point of this calculator.
The calculator itself is model-based. Historical calibration and score-frequency evidence are reported separately so retrospective results are not presented as guarantees for a new match.
Market accuracy
The separate market-calibration study covers 19,381 matches across five European leagues and reports league- and probability-cluster differences between de-vigged market probabilities and observed outcomes. Those are retrospective diagnostics, not fixed market-bias rules. Open the calibration study →
The models
Poisson
Independent goals, fixed rate λ per team
Dixon-Coles
Corrects 0:0, 1:0, 0:1, 1:1 via ρ
FAQ
Does this calculator predict the correct score?
No. It shows the probability distribution implied by the market. Even the most likely exact score is often only a low-double-digit probability in typical football markets; the exact value depends on the supplied odds and fitted model. This is a visualization tool, not a prediction engine.
Can I calculate football win probability with this tool?
Yes. The same pre-match score matrix gives Home, Draw and Away probabilities by summing the relevant score cells. It covers association football (soccer). The displayed model probabilities depend on the supplied odds and model settings; they are not a live, in-play win-probability feed.
Where do I find the odds to enter?
Use complete 1X2 and, if available, Over/Under 2.5 markets from one bookmaker at the same pre-match timestamp. An odds comparison source can help locate those prices, but mixing best prices from different bookmakers creates a synthetic market. For historical analysis, retain the source and timestamp.
Which model should I use — Poisson or Dixon-Coles?
Use Poisson as the transparent independent baseline. Dixon-Coles applies a local correction to 0:0, 1:0, 0:1 and 1:1. A non-zero ρ should be estimated, externally validated or used explicitly as a sensitivity assumption.
Can I use this for any league?
The calculator can process valid football market odds from any league, but model calibration and Dixon-Coles ρ are not automatically transferable between competitions. Treat league-level historical values as diagnostics rather than universal constants.
What does "Model Fit" show?
It shows how closely the fitted score model reproduces the de-vigged market targets. Smaller residuals mean a closer numerical fit. They do not tell you whether the probabilities will be well calibrated on future matches.
Can this tool help me win bets?
It is not designed to provide a betting edge. It translates market prices into a coherent probability view. Calibration and historical ROI are separate empirical questions, discussed in the market-calibration study linked above.
Is my data stored?
No. All calculations run in your browser. Nothing is sent to any server. The Share URL encodes inputs in the link itself.
How is the bookmaker margin removed?
The calculator currently uses proportional normalization: raw implied probabilities are divided by the market booksum so the de-vigged probabilities sum to 100%. Alternative margin-removal methods are compared in the No-Vig Calculator linked above.