Full scoreline probability matrices, 1X2 percentages and expected goals for today's fixtures — generated by our Poisson distribution model calibrated with 38-match rolling averages and Elo ratings. No account required.
Every mathematical prediction on this page is free to access. Read the full model methodology.
Every prediction on this page uses a Poisson distribution model calibrated with 38-match rolling averages, adjusted for home advantage, opponent Elo strength, and form weighting. The model outputs a full scoreline probability matrix, 1X2 probabilities, and expected goals for each fixture. Predictions refresh daily by 08:00 UTC. Read the full methodology.
Showing 20 of 326 fixtures
The xGaura mathematical predictions engine uses the Poisson distribution — a statistical model that calculates the probability of a given number of events occurring in a fixed interval, given an average rate of occurrence. Applied to football, it answers the question: if Team A scores an average of 1.8 goals per home game and Team B concedes an average of 1.2 goals per away game, what is the probability of Team A scoring 0, 1, 2, 3 or more goals in this specific fixture? Repeat the calculation for both sides, combine the output into a full scoreline matrix, and you have a complete probabilistic picture of the match.
A Poisson model treats goals as independent events occurring at a constant average rate. That is a simplification — goals are not strictly independent, and rates do shift within matches — but across a large sample of football matches, the Poisson approximation is remarkably accurate. It correctly predicts the most likely scoreline in top-flight European football around 13–16% of the time, and correctly predicts 1X2 outcomes in roughly 50–55% of cases when calibrated properly.
The key to a good Poisson model is the quality of the inputs. Raw goals scored and conceded averages are a starting point, but they are noisy. xGaura uses expected goals (xG) as the primary input rather than actual goals, because xG is a better predictor of future scoring than past scoring alone. A team that has scored 8 goals from 2.1 xG over five games is likely overperforming and will regress; a team that has scored 2 goals from 3.8 xG is likely underperforming and should improve. Our model captures this.
The raw Poisson output is adjusted by the Elo rating differential between the two teams. Elo ratings are a measure of team strength derived from historical results, updated after every match based on the outcome and the difficulty of the opponent. A team with a significantly higher Elo rating than their opponent will have their scoring expectation scaled up and their concession rate scaled down, and vice versa. This adjustment prevents the model from treating every match as if the two sides are of equal quality, which raw xG averages can obscure in small samples.
Each prediction card on this page shows the four most likely scorelines for that fixture, along with each scoreline's individual probability. The hot cell (highlighted in blue) is the single most likely exact scoreline according to the model. The warm cells are the second and third most likely. These are not tips — they are model outputs. The probability of any individual exact scoreline is inherently low, even the most likely one. For correct score tips backed by model edge, see the Correct Scores page.
The xGE (Expected Goals Estimate) figure shown in each card is the model's prediction of how many goals each team is expected to score in this specific fixture, after all adjustments for opponent strength, home advantage and Elo. It is not the same as a team's season xG average — it is a match-specific number. A high xGE for one side and a low xGE for the other strongly indicates a likely Over result on the goals line. A low combined xGE points to an Under. These figures feed directly into the Over/Under predictions on this site.
xGaura's Poisson model correctly predicts the 1X2 outcome in approximately 78% of tested fixtures when selecting the highest-probability outcome. However, that figure should be interpreted carefully. The model is not predicting 78% of matches correctly by picking a side — it is selecting the outcome its probability distribution favours most, and verifying accuracy against that. In genuinely 50/50 fixtures, the model will be correct roughly half the time. Its real advantage is in identifying fixtures where the bookmaker's implied probability significantly undervalues one outcome — the foundation of value betting.
For the complete mathematical derivation of the model, including the Poisson formula, the Elo adjustment function and the home advantage coefficient, see the methodology page. Bet responsibly. 18+ only.