Methodology
How KickSignal
predicts matches
We believe in transparency. Here's exactly where our data comes from, how the AI turns it into probabilities and scenarios, what those numbers mean — and their limits. No black box, no magic promises.
Data
Our data sources
Official match data
Match data is sourced from reliable providers. The AI model builds its analysis exclusively on verified historical match data from the active data provider.
xG & advanced stats
xG performance over the last eight matches is included for teams with sufficient data coverage. Shots, shots on target, possession and corners — all critical match statistics — are evaluated together to produce more accurate predictions.
Rest period and player absences
Rest time between matches and league performance are factored into the analysis. Injured and suspended players are weighted by their importance to the team. Head-to-head history (H2H) and starting line-ups are also evaluated by the model to contribute to the analysis.
Real-time and transparent analysis
Fixtures and match results are updated regularly, while live scores are refreshed at shorter intervals. Every user analysis is a timestamped snapshot created at the moment of request. Public AI predictions are recorded before the match starts and are never altered afterwards.
Calculation
How the model calculates
- 01
Deterministic analysis model
Probabilities are calculated using a time-weighted custom analysis model. Missing numerical data is not interpreted as poor performance — it defaults to neutral. The AI-generated written match analysis is produced based on the numerical prediction model's results.
- 02
Calibrated, coherent probabilities
The win / draw / loss distribution is calibrated with a temperature selected on the validation period and remains mathematically coherent: it sums to 100%.
- 03
One score distribution
1X2 probabilities, the primary score scenario, alternatives, Over 2.5 and BTTS all come from the same goal distribution. Bookmaker prices are not blended into serving output.
How to read it
What the numbers mean
- 1X2 probabilities
- An estimate of each outcome's relative likelihood. A team with a 60% win probability is not certain to win; it should fail to win roughly four times in ten under similar conditions.
- xG (expected goals)
- Observed xG from previous matches is a form signal. The expected-goals figure on a prediction card is the model's projected scoring average for the future fixture; they are not the same measurement.
- Scenarios
- The primary score is the highest-probability score scenario consistent with the leading 1X2 outcome. Alternatives express uncertainty; none is a guaranteed result.
Limits & honesty
Football is unpredictable by nature: a red card, a penalty, a mistake can change everything. Our predictions are reasoned estimates based on available data, not certainties. Quality depends on data freshness and completeness; on lightly covered competitions some stats may be missing (and are not invented).
Not betting advice
KickSignal is an AI-powered statistical analysis platform. The predictions and analysis provided are for informational purposes only and do not constitute investment, betting or earnings advice, incentive or guarantee.