How the numbers are built.
StatsXI is designed to make the denominator visible. A percentage is only useful when you can see the sample behind it, and a historical comparison is only useful when you understand how that sample was selected.
Sample-aware statistics
Every average or rate keeps its own observed sample. If corners are missing for one match but the final score is present, that match can still contribute to goal statistics without being counted in the corner denominator.
Missing source values are treated as missing, not as zero.
Team windows and recent form
Team profiles can be calculated over recent windows such as Last 5, Last 10, Last 20 or the selected season. Context filters can separate overall, home and away performance where the underlying data supports it.
For an upcoming fixture, recent-form metrics use matches from the same league and season that occurred before the selected fixture, preventing future matches from leaking into the calculation.
Market-implied probabilities
Where informational pre-match odds are available, StatsXI converts the 1-X-2 market into implied probabilities and mathematically normalizes them to remove the observed market margin before comparison. The resulting profile is a descriptive representation of how the historical market priced the match.
A normalized probability is not a StatsXI forecast and is not an assessment of "fair odds", expected return, betting value or an advantageous opportunity. The transformation is used only to make market snapshots easier to compare statistically.
Historical match similarity
Similarity analysis searches historical matches with market profiles close to the selected fixture and widens the matching tolerance through quality tiers when necessary. The final sample is then summarized using observed outcomes, goal patterns and other available statistics.
Percentages are withheld when the comparable sample is below the minimum required threshold. Where interval estimates are shown, StatsXI uses Wilson 95% confidence intervals.
Similarity to past matches is descriptive and does not imply that the selected fixture will produce the same outcome.
Half-time scenarios
The half-time explorer conditions historical matches on an exact interval score such as 0-0 or 1-0, then summarizes what happened by full time and during the remaining period. It does not assume that full-time event totals happened after half time when the source data does not provide event timestamps.