Foresportia FAQ
Probabilities, reliability, markets, data, leagues, history and API: detailed answers to understand how to read Foresportia today.
The core principle: a prediction is not a certainty. The right question is: how probable, robust and context-consistent is this scenario?
Choose your topic
Jump directly to the subject you need.
Probabilities: how to read percentages
Understand what 55%, 60% or 70% really means — and why a high probability can still fail.
What does a 60% football probability mean?
A 60% probability means that across a large number of comparable situations, the scenario is expected roughly 6 times out of 10.
For one isolated match, it never means “this will happen”. The correct reading is: this is the most likely scenario, with 40% uncertainty remaining.
Why is probability ≠ certainty especially important in football?
Football is a low-scoring, high-variance sport. A penalty, red card, injury, individual error or deflection can dramatically alter the outcome.
Probability quantifies uncertainty; it does not eliminate it.
Why can a 70% probability fail several times in a row?
Because 70% still implies roughly 30% theoretical failure. Over short runs, randomness can create clustered misses without invalidating the model’s long-run calibration.
Is a higher probability always better?
Not automatically. A high probability is useful if it is well calibrated, supported by enough data and consistent with the competition context.
Foresportia always reads probability + robustness + context.
What is the difference between 1X2 probability and exact score?
1X2 distributes probability across three broad outcomes: home, draw and away. Exact score spreads probability across many more possible results.
Exact score is therefore naturally more dispersed and fragile.
Reliability, calibration and history
A good AI model is not judged only by how many results it gets right: its probabilities must also match reality over time.
What is calibration?
Calibration asks a simple question: when the model says 60%, do we observe about 60% success over enough comparable cases?
A model can rank matches well while still outputting poor percentages. Calibration aims to make probabilities honest.
What is a reliability curve?
It groups predictions into probability bins and compares announced probability with observed frequency.
- close to the diagonal: good calibration;
- above: the model is rather conservative;
- below: the model may be overconfident.
What are Brier Score and LogLoss used for?
Brier Score measures the gap between predicted probabilities and outcomes. LogLoss strongly penalizes highly confident mistakes.
They are more informative than a raw hit rate alone.
What does confidence or stability mean?
It complements raw probability with a robustness layer: historical volume, league behaviour, signal stability and scenario consistency.
Two matches at 65% may therefore not deserve the same interpretation.
Where can I verify Foresportia performance?
The Past results page shows settled results and historical performance. For derivative markets, use the available picks history as well.
55%, 60%, 70% thresholds: coverage vs precision
Raising a threshold reduces the number of matches displayed. That can increase selectivity but also shrink the sample.
Why is 55% often used as a starting point?
Because it removes part of the near coin-flip population while preserving enough volume to analyse results.
It is not a universal threshold.
Why can a 70% threshold be misleading?
Because it can drastically reduce sample size. On a small sample, a very good or very bad run can mostly reflect variance.
Always read accuracy together with number of observations.
Is there a universal best threshold?
No. It depends on competition, season maturity, data volume and the trade-off you want between coverage and selectivity.
Why does performance change when I change the threshold?
Because you are changing the population being analysed. At 55% you include more matches; at 70% you keep only the strongest model signals.
1X2, BTTS, Over/Under, DNB, double chance and exact score
Each market answers a different question. They should not be read as interchangeable versions of the same prediction.
What do 1X2, BTTS, Over/Under, DNB and double chance mean?
- 1X2: home / draw / away.
- BTTS: both teams to score.
- Over/Under: total goals above or below a threshold.
- DNB: Draw No Bet.
- Double chance: combines two 1X2 outcomes.
Why does Foresportia show several markets for one match?
A match may be uncertain on the winner but clearer on goals — or the opposite. Cross-reading markets helps reveal the probabilistic structure of the fixture.
How should I read exact scores?
As a distribution of scenarios, not as a certain score. Foresportia presents several likely outcomes and relates them to BTTS and Over/Under context.
Why is exact score harder than 1X2?
Because probability is spread across many more outcomes. Even the top exact score can have a relatively low probability.
Are derivative markets always published?
No. Depending on data quality and quantity, some markets may be hidden. Foresportia prefers not to publish rather than display a signal judged too fragile.
Leagues, countries and season maturity
Not every competition has the same variance, historical depth or statistical stability.
Why does the same probability not mean the same thing in every league?
Competitions differ in draw rate, goals per match, team parity, dominance structure, variance and historical depth.
Why do some leagues look more predictable?
When team-strength gaps are more stable and data is deeper, some structures are easier to model. Surprises never disappear.
Why are early-season matches more difficult?
Promotions, relegations, transfers, new coaches and tiny samples make early rounds more uncertain. Foresportia uses safeguards and may restrict publication.
Why are some matches missing or never published?
A match can be excluded when data is insufficient, inconsistent, immature or too fragile for a public prediction.
Does Foresportia only cover major European leagues?
No. The catalogue also includes secondary leagues and competitions outside Europe, depending on data availability and quality.
Which Foresportia page should I use?
Each page has a distinct role. This section helps you choose the right entry point.
I want today’s best selections: where should I go?
Use Top AI predictions. It highlights the strongest public signals across available markets.
I want to browse every match on a date: where should I go?
Use Matches by date. It is the most convenient page for scanning an entire matchday.
I want to analyse one specific match: where should I go?
Use the Direct hub, designed for a guided match-by-match reading.
What is the difference between football predictions and prediction modelling pages?
The predictions page is more practical and selection-oriented, while the modelling-oriented prediction page focuses more on probabilities, structure and interpretation.
Where can I see team form and ELO?
Team pages and Team Form Insights group form, streaks and other contextual indicators.
Where can I understand the full methodology?
Start with AI methodology, then compare the principles with historical results.
Data, signals, ELO and context
Foresportia prioritizes measurable, reproducible signals that can be integrated reliably into an automated pipeline.
What kinds of signals does Foresportia use?
The system combines attack/defence strength, team dynamics, home/away effects, competition context, ELO and goal distributions depending on the market and data availability.
Does the model use injuries, motivation and line-ups?
Very fresh or hard-to-objectify information can be difficult to integrate robustly. Foresportia prioritizes auditable data and recommends human context checks when needed.
What is ELO used for?
ELO provides a compact signal of relative team strength. It does not replace other variables but complements them when reliable enough.
Why can a promoted club’s history be less useful?
Because previous matches were played in a different competitive environment. Opponent quality and result distributions change after promotion or relegation.
Are Foresportia data and predictions updated automatically?
Yes. A large part of the pipeline is automated: fixtures, results, calculations, enrichments and publication are regenerated regularly.
Drift, bias, common mistakes and responsible reading
A model can be good and still degrade, go through bad runs or become less adapted to some contexts.
What is drift in football prediction?
Drift is the gradual change in distributions: playing styles, refereeing, league pace, transfers, promotions, rules or tactical trends.
A model must be monitored and re-evaluated over time.
Can I judge the model on 10 or 20 matches?
No. A sample that small is too sensitive to randomness. Evaluation requires sufficient volume, ideally by league, market and threshold.
Why can a strong period be followed by a weak one?
Even a calibrated model experiences adverse runs. Variance, drift, seasonality and context changes can temporarily affect results.
What is the #1 mistake when reading a prediction?
Turning a probability into a statement: “this will happen”. The correct reading remains: how probable, robust and context-consistent is it?
Does Foresportia promise profits or sure picks?
No. Foresportia is a probabilistic analysis tool. It does not guarantee outcomes and does not market predictions as certainties.
Free access, account, advertising, app and API
Practical answers about access to the website and the different Foresportia products.
Is Foresportia free?
Yes. The main public prediction, statistics and history pages are available for free. Advertising helps fund the project.
Do I need an account to view predictions?
No for the main public pages. An account may provide additional options depending on the available offer.
Why is there advertising on Foresportia?
Advertising helps fund infrastructure, data and product development while keeping broad public access free.
Is there a Foresportia app?
Yes. Foresportia is also available as an app to make match analysis easier to access on mobile.
Does Foresportia provide an API?
Yes. The Foresportia API provides structured data and predictions for developer and product use cases.
Can I report a bug or suggest a feature?
Yes. Use the Contact page and choose bug, question, suggestion or partnership.
Mostly looking for the right page to use?
The free hub points you to the right tool depending on your need: today’s shortlist, one specific match, history, teams, statistics or AI methodology.