Football Results Predictor - Elo + xG + Poisson
Enter the Elo ratings and expected goals (xG) for both teams to estimate the 1X2 odds (home win, draw, away win) using the Poisson statistical model.
How to predict football match results
The prediction model combines two established statistical approaches: the rating Elo (Árpád Élő, 1978) to measure the relative strength of teams, and the Poisson distribution (Maher, 1982) to estimate the probability of every possible score. The difference in Elo between the two teams determines the proportion of expected goals (lambda) assigned to each; the Poisson distribution translates these lambdas into a matrix of probabilities for every single outcome up to 8-8.
If you have the Expected Goals (xG) data from recent matches, entering them significantly improves accuracy: xG capture the quality of the chances created, not just the goals actually scored, reducing statistical noise. Poisson+Elo models typically achieve an accuracy of 50–55% on predicting 1X2 outcomes (compared to ~46% for a random choice).
Configure
Result
No results
Enter values and press the button to calculate.
Frequently Asked Questions
How does the Elo model work for soccer?
The Elo rating is a relative strength ranking system originally developed for chess by Árpád Élő (1978) and adopted by football. A difference of 200 points corresponds to approximately a 76% chance of victory for the stronger team. The model combines Elo with the Poisson distribution to estimate expected goals per team.
What are expected goals (xG)?
Expected Goals (xG) measure the quality of scoring opportunities based on shot position, distance from the goal, and type of assist. An xG of 1.5 means these chances should have resulted in an average of 1.5 goals. Adding them increases prediction accuracy beyond Elo rating alone.
How is the Poisson distribution used?
The Poisson distribution models rare and independent events with a known average rate - just like goals in soccer (Maher 1982). Given the lambda for each team, we calculate the probability of every possible score up to 8-8. Summing the matrix probabilities gives us the 1X2 odds.
How accurate is this predictor?
Poisson + Elo models typically achieve 50-55% accuracy in predicting the 1X2 outcome (vs ~46% randomly). The model doesn't account for injuries, lineups, weather or motivation. For informational and recreational use - not financial decisions.
Where do I find the Elo ratings for football teams?
Club team Elo ratings are available at clubelo.com (with history and free API), footballdatabase.eu, and various stats sites. For national teams, FIFA publishes its own ranking. Soccer Reference features advanced Elo variants like FiveThirtyEight's SPI.
Is the sum of the 1X2 probabilities always 100%?
With the truncated Poisson model at 8 goals per team, the sum is slightly less than 100% (typically 99.8–99.9%) because extremely unlikely outcomes like 9-0 or 10-0 are ignored. The tool displays the actual sum for transparency.
How is it used?
- Enter Elo ratings
The Elo rating measures a team's relative strength. Italian Serie A teams typically have ratings between 1400 and 1900. You can find current ratings on sites like clubelo.com or footballdatabase.eu.
- Enter expected goals (xG) - optional
Expected goals (xG) represent the quality of scoring chances created by each team in recent matches. If available, they replace the pure Elo model estimate for a more accurate prediction.
- Calculate probabilities
The model applies the Poisson distribution to estimate the probability of each possible outcome (0-0, 1-0, 0-1, etc.). The 1X2 probabilities (home win, draw, away win) are derived from the resulting matrix.
- Interpret the results
The 1X2 odds indicate how likely each outcome is according to the model. An AI analysis (Variant A) is available after calculation to contextualize the results with qualitative factors not captured by the statistical model.
Glossary
- Elo Rating
- A system for rating the relative strength of two players/teams based on historical results. A 200-point difference → ~76% win probability for the stronger team.
- Expected Value of the Goal
- Metric that quantifies the quality of scoring opportunities based on position, distance, and type of shot. Value between 0 and 1 for each chance; sum for the entire match = total xG.
- Anonymous function
- Poisson distribution parameter representing the expected average rate of events (goals) in a fixed interval. Derived from Elo or provided directly as xG.
- Poisson Distribution
- Discrete probability distribution that models the number of independent events with a known average rate in an interval. Applied to soccer by Maher (1982) to estimate the probability of each score.
- 1X2 Odds
- European betting system: 1 = home team win, X = draw, 2 = away team win. The three probabilities sum to ~1 (model truncated at 8 goals per team).