Riftbound / Insights
Insights
What the analysis layer will look like — built now so the standards are visible before any real number exists.
Future capability · demonstration data only. Nothing on this page is measured. It is an interface preview showing exactly how figures will be qualified when there are figures to publish.
Matchup performance
DemoMatchupWin ratenConfidencePlay / draw
vs Kennen54.2%96Moderate+3.1pp on the play
vs Master Yi51.8%71Low+1.4pp on the play
vs Irelia47.5%118Moderate−4.9pp on the draw
vs LeBlanc mirror50.0%44Insufficient—
Demonstration data · deck v1 · Aug 2026 snapshot · source: none (interface preview)
Insufficient data
LeBlanc mirror · n = 44. Below the reporting threshold.
Shown as a state, not hidden, and never extrapolated. A thin sample that looks like a 50% matchup is not evidence of a 50% matchup.
Pilot patterns · planned
- Scoring windows missed through own conversion
- Overconversion frequency
- Sideboard dilution below the 26-unit floor
- Power-management errors before Herald deaths
- Repeated sequencing errors by taxonomy tag
Statistical integrity
- Sample size shown on every statistic
- Confidence ranges, never bare point estimates
- Data freshness and meta-snapshot dates
- Deck-version and matchup context
- Play/draw split where it matters
- Correlation flagged as correlation
- Insufficient-data states instead of thin claims
- Demonstration data always labeled