Predictions

Football predictions

Win probability, BTTS and over/under models for each match. Based on real data.

The full predictions are available at the Starter plan and above.

Football predictions for today

Loading matches…

AI predictions · LigaToday

Predictions you can rely on

The football predictions on LigaToday are algorithmically calculated probabilities for the outcome of every match – 1X2, whether both teams score (BTTS) and over/under 2.5 goals – across 680+ competitions, free and with no registration. Two models sit behind them: a free ELO rating paired with a Poisson goals model (~52% accuracy on the 1X2 market in the top 5 European leagues) and the Pro XGBoost model with 23 features (~55% accuracy), backtested over 8 seasons of historical data from our licensed sports-data providers. Predictions are recalculated every hour, so you always see the current probabilities for today’s matches. If you’d prefer to follow a specific analyst rather than a pure model, take a look at our section on tipster predictions.

  • 1X2 odds

    Actual odds for home win, draw and away win – calculated based on form, xG and head-to-head matches. Free for every match.

  • Both Teams to Score + Over/Under 2.5

    How likely is it that both teams will score and that more than 2.5 goals will be scored? Based on the season’s average number of goals.

  • Expected goals (xG)

    An xG model for each match shows which team is expected to dominate in terms of the quality of chances created, rather than simply in terms of strength.

  • Pro AI with XGBoost

    23 variables, overnight retraining, results validated back to 2018. Pro predictions with full justification.

How AI predictions work

  1. 1

    Data collection

    We extract ~120 statistics per match - result, xG, shots on target, corners, cards, possession, starting line-ups, home/away advantage and the break between matches.

  2. 2

    ELO + Poisson (free model)

    The ELO rating indicates the relative strength between the two teams. The Poisson model converts expected goals into a distribution of results → probabilities for 1X2, BTTS and over/under 2.5.

  3. 3

    XGBoost (Pro model)

    Machine learning with 23 variables, analysed every night. This takes into account form, head-to-head records, injured players, weather conditions and odds movements. Result: ~55% accuracy 1X2 in the top five leagues.

  4. 4

    Backtesting

    Each model is tested against eight seasons’ worth of historical data (2018–2026) before going into production. The backtest results are available at /predictions/methodology/.

Glossary of markets

What do the abbreviations you see in our predictions mean?

1X2
The three possible outcomes of the match: 1 = home win, X = draw, 2 = away win. Our prediction gives the probability of each outcome as a percentage.

BTTS
Both Teams to Score – both teams score. „Yes“ if both score at least one goal, „No“ if either team fails to score.

Over/under 2.5
Total number of goals in the match. “Over 2.5” means 3 or more goals; “Under 2.5” means 0–2 goals. A classic market for high-scoring or low-scoring matches.

xG (expected goals)
The sum of the expected goals (xG) for all goal-scoring opportunities in a match. A team with 2.4 xG against 0.6 xG has dominated, even if the score is 1–1.

Form
The team’s last 5–10 matches, weighted according to the strength of the opposition. These are given greater weight in the model than results from 3+ months ago.

Head-to-head (H2H) matches
A look back at the last 10 matches between the two teams – goals, clean sheets and wins. Not very predictive, but useful for context.

Frequently Asked Questions

What are football predictions?

Football predictions are an assessment of the likely outcomes of a match, expressed as a percentage. On LigaToday, they are generated automatically by statistical models and cover the 1X2 market (home win, draw, away win), whether both teams will score (BTTS) and over/under 2.5 goals. These predictions are for guidance only and do not guarantee the result.

Are the AI predictions on LigaToday free?

Yes. The 1X2, BTTS and over/under 2.5 odds for each match are completely free and do not require registration. The Pro AI prediction (XGBoost + 23 variables + rationale) requires a Starter plan or higher.

How accurate are the predictions?

The free ELO + Poisson model achieves an accuracy of ~52% for the 1X2 market in the top 5 European leagues (compared to 33% for random predictions). The Pro AI XGBoost model achieves ~55% and has been back-tested to the 2018/19 season. The full metrics are available at /predictions/methodology/.

When are the predictions updated?

Predictions are automatically updated every hour, and the Pro model is retrained every night once all the day’s matches have finished. Starting line-ups, announced approximately 60 minutes before kick-off, are not included in the free version, but are included in the Pro version.

Where does the data come from?

Results, statistics and xG come from our official sports-data partners, covering 680+ competitions. Top scorers, head-to-head records and form are calculated directly from historical fixtures in our own database.

Can I follow only certain leagues?

Yes. On each league’s page (e.g. /league/champions-league-europe/), you’ll see predictions for that league’s matches only. With a free account, you can bookmark up to 10 favourite leagues and receive pre-match notifications.

Where can I find the predictions for the matches being played today?

All the fixtures for the day are listed on the page today’s matches, and the calculated probabilities for them – in football predictions for today. As the models are recalculated every hour, the probabilities remain up to date until the start of each match.

What is the difference between model predictions and tipster predictions?

The model predictions are calculated entirely by algorithms (ELO + Poisson and Pro XGBoost), without any human intervention. In the section on tipster predictions The focus is on analysis by a specific author, who can provide additional context such as injuries, motivation and form. The two approaches complement each other – we recommend that you compare the model’s probability with a particular tipster’s perspective before forming your own opinion.