Independent football data and predictions, worldwide.
LigaToday is an independent international platform for live scores, predictions and full statistics across 680+ leagues in 74 countries - 8,900+ teams, 66,000+ players and 3,900+ stadiums. One owner, one open model, no betting partnerships.
About LigaToday
An independent international football site, founded in 2025 by Simeon Kirilov. No betting partnerships, no hidden fees, no editorial bias - just data, methodology and a published accuracy record.
Our mission
For years, football fans have had to choose between a slow-loading SofaScore, aggregators cluttered with adverts, or threads on niche forums. LigaToday is the alternative: fast, free for fans, and with a fully transparent methodology behind every published prediction.
Our aim is simple - to publish true probabilities, rather than “tips”. We don’t say “bet on X”. We say “the model estimates a home win at 56% with a Brier score of 0.58 compared to a baseline of 0.667 for a uniform prior“ - and we leave it up to you to decide whether 56% is sufficient.”.
Who is behind LigaToday?
Simeon Kirilov - a software engineer with over 10 years’
experience in data-driven products. Former specialist in the iGaming industry
(hence the username @iGamingKirilov), now focusing
on transparent statistics that the industry itself does not provide.
Simeon is the sole author of:
- ELO + the Poisson model and the code behind it (PHP, WordPress, MySQL).
- Pro AI XGBoost model (Python), 23 pre-match signals, 5-fold stratified cross-validation.
- LigaToday REST API (
/wp-json/sm/v1/). - The methodology page, which documents every step of the model.
Link: kirilov@seolinked.com · LinkedIn · Telegram @iGamingKirilov.
How LigaToday works
Live scores are updated every 30 seconds. A REST endpoint powers the live page, the fixtures page and the built-in widgets for each match.
The predictions are generated using two parallel models:
- Free model - ELO + Poisson. Each team has a numerical ELO rating (K=20, home bonus +60). After each match, the ratings are updated according to the standard ELO logistic curve within ~60 seconds. The expected goals are Poisson-distributed, with the mean goals-per-match for the league season as the base. The convolution of the two Poisson distributions yields a complete probability matrix for every possible result. From this, we derive 1X2, Both Teams to Score (BTTS), over/under 2.5 and the most likely correct score.
- Pro AI model - XGBoost. Gradient-boosted decision trees (500 trees, depth 3, learning rate 0.04, L2 regularisation 2.0). Three separate models for 1×2, BTTS and over/under 2.5. 23 pre-match signals: ELO difference, form over the last 5 and 10 matches, goals for/against, form by position, days’ rest, head-to-head records, etc. Validated with 5-fold stratified cross-validation. We retrain the model every night at 03:30 UTC.
The full formulae, K-factors, validation and data sources are documented in the methodology page. This is a public commitment - the figures we present before the match are the same ones against which we are assessed after the match.
Transparency
We do not promote sports betting operators. There are no affiliate links, no sponsored placements and no commercial partnerships with bookmakers anywhere on this site. Where bookmaker odds appear, they are published as market data next to our own model probabilities, and we earn nothing from them.
- No betting partnerships. We do not carry affiliate links to bookmakers and take no commission from them, so nothing commercial can influence the data, the predictions or the standings.
- We do not edit historical predictions. The figure from the day before the match is the one we use to assess ourselves.
- Open methodology. The complete algorithm, all K-factors and hyperparameters have been published at /predictions/methodology/.
- Open REST endpoint Regarding the record:
/wp-json/sm/v1/predictions/accuracy. You can download it and run the numbers yourself. - One team, one owner. There is no anonymous board, no hidden investor. Simeon Kirilov responds in writing to every audit.
What can you do on LigaToday?
Frequently asked questions
- What is LigaToday?
- LigaToday is an independent international football data platform. We cover 680+ leagues in 74 countries, 8,900+ clubs and national teams, 66,000+ players and 3,900+ stadiums. We publish live scores (refreshed every 30 seconds), full match statistics, standings, top scorers and our own match predictions.
- Who is behind LigaToday?
- The platform was built and is maintained by Simeon Kirilov, a software engineer with over 10 years of experience in data-driven products. He is the author of the ELO + Poisson model and the Pro AI XGBoost model, of the codebase, the methodology and this page. No owner or investor dictates the content.
- How does LigaToday make predictions?
- We run two models in parallel. The free one combines ELO ratings (K-factor 20, home bonus +60) with a Poisson distribution for goals. Pro AI is XGBoost with 23 pre-match signals, 5-fold stratified cross-validation and hyperparameters of 500 trees / depth 3 / learning rate 0.04 / L2 = 2.0. We retrain nightly at 03:30 UTC. Full methodology: /predictions/methodology/.
- Are the predictions free?
- Yes. Live scores, match statistics, standings, top scorers, the free ELO + Poisson prediction and the pre-match briefings are all free and need no sign-up. The Pro AI prediction (the XGBoost model) and the additional signals are part of the Pro plan.
- Where does the data come from?
- A licensed football feed for the primary data (scores, events, line-ups, statistics), plus our own computation for ratings, form, head-to-head records and predictions. All of the maths runs on our own servers – there is no third-party prediction service, no bookmaker feed and no closed „AI“.
- Do you have partnerships with betting operators?
- No. We have no partnerships with betting operators, no affiliate links and no sponsored placements. We publish bookmaker odds only as market data alongside our own model probabilities, and we receive no commission from any operator.
- How is accuracy measured?
- Every finished prediction is frozen and scored against the real result. We publish top-pick accuracy (the share of matches where the highest-probability outcome matched reality), the Brier score (the squared distance between the probability and the actual binary outcome) and the sample size. We never edit historical predictions. The raw data is available at
/wp-json/sm/v1/predictions/accuracy. - How often are the results updated?
- Live scores refresh every 30 seconds. ELO ratings are recalculated within about 60 seconds of the final whistle. Each league goal environment (the Poisson baseline) is recalculated daily. The Pro AI XGBoost model is retrained nightly at 03:30 UTC.
- How can I get in touch with the team?
- Email: kirilov@seolinked.com. LinkedIn: mt.linkedin.com/in/simeon-kirilov. Telegram: @iGamingKirilov. We reply in writing to every audit of a calculation and every content complaint.
Ready to get started?
Log in to your free account and start following your favourite teams and the predictions for them.