footballtoday
54′ United Arab Emirates crestUnited Arab Emirates2 Honduras crestHonduras1 HT Saudi Arabia crestSaudi Arabia1 South Korea crestSouth Korea0 39′ Qatar crestQatar0 Australia crestAustralia2 FT Llaneros crestLlaneros0 Chico crestChico0 FT Czechia U18 crestCzechia U180 Poland U18 crestPoland U183 FT Singapore crestSingapore1 Paraguay crestParaguay4 12:00 Chinese Taipei crestChinese Taipei Kyrgyzstan crestKyrgyzstan FT Comoros crestComoros0 Congo crestCongo0 FT Mozambique crestMozambique0 Ethiopia crestEthiopia1 FT Zambia crestZambia1 Burundi crestBurundi1 13:00 Congo DR U23 crestCongo DR U23 Nigeria U23 crestNigeria U23 16:00 Burkina Faso crestBurkina Faso Mauritania crestMauritania 16:00 Cape Verde Islands crestCape Verde Islands Liberia crestLiberia 16:00 Gabon crestGabon Lesotho crestLesotho 16:00 Nigeria crestNigeria Tanzania crestTanzania 16:00 South Africa crestSouth Africa Kenya crestKenya 16:00 Tunisia crestTunisia Libya crestLibya 16:30 Kuwait crestKuwait Syria crestSyria 16:30 Poland U21 crestPoland U21 Norway U21 crestNorway U21 17:00 Moldova crestMoldova Kazakhstan crestKazakhstan 17:00 JS Saoura crestJS Saoura Khenchela crestKhenchela 17:00 Germany U21 crestGermany U21 Belgium U21 crestBelgium U21 17:30 France U21 crestFrance U21 Italy U21 crestItaly U21 17:30 Chile U20 crestChile U20 Mexico U20 crestMexico U20
Transparency

How our numbers work

The maths behind match predictions, season simulations, fixture difficulty and the texts on our match pages.

1. The match model

We use a Poisson goals model. For every club we measure four ratings from the current league table: attack and defence at home, and attack and defence away, each relative to the league average.

Expected goals for the home side = league average home goals × home attack rating × away defence rating of the opponent (and the same in reverse for the away side). Early in the season every rating is blended with four "average" matches so that a couple of freak results do not dominate.

From the two expected-goals figures we compute the probability of every scoreline from 0–0 to 10–10, which gives the home win, draw and away win percentages and the most likely score.

For cup ties between different divisions, each club is rated inside its own league and a division-strength factor is applied. Finals are treated as neutral venues.

2. The season simulator

We play every remaining league fixture 10,000 times using the match model, add the simulated points to the real table, rank clubs by points, goal difference and goals scored, and count how often each club finishes in each position.

Scottish Premiership: we simulate up to the split after round 33, divide the table into top and bottom six, and — if the post-split fixtures have not been published — generate the five remaining rounds inside each half. Clubs cannot move between halves after the split, exactly as in the real competition.

3. Fixture difficulty

Each upcoming opponent is rated from 1 (easiest) to 5 (hardest) according to its current league position, with half a point added for away matches.

4. Match texts

Previews and reports are written from facts computed by our code: positions, points, form, head-to-head records and model probabilities. Where an AI writing assistant is used, it receives only those facts, and every number in its text is automatically checked against them. Texts that fail the check are discarded and replaced by our template text. Betting vocabulary is blocked.

5. What the model doesn't know

Probabilities describe uncertainty; they are not guarantees. Football Today does not provide betting advice.

Data

Fixtures, scores, line-ups, statistics and tables come from our football data provider and are refreshed continuously — every 15 seconds for live matches.