Reproducible One Piece Card Game research
Evidence for the matchups players actually debate.
TCG AI SIM runs complete games between rated decks, then publishes the sample, uncertainty and exact evidence cut behind every result. Explore measurable AI simulations—not another unsourced tier list.
Simulated evidence complements tournament results; it does not predict every pilot, local meta or future set.
Current field evidence
Leaders with the strongest measured records
Ranked here by 95% Wilson lower bound across decisive Standard RATING appearances. The deck-level ladder uses a different Glicko rating and lives on the ranking page.
Matchup laboratory
Where the field looks most asymmetric
Only pairings above the 20-game publication floor are shown. Every result links to both leader evidence hubs and retains its confidence interval.
Play vs draw
Does going first matter?
Across 12,961 games with a recorded opener, the first player won 50.2%.
95% CI 49.4%–51.1%Follow the evidence
Explore by question, not by content type
Which leaders and decks rate highest?
Glicko ratings, records, uncertainty and drill-downs into every public deck.
Explore →MetagameWhich sets contribute the measured field?
Current Standard leader cohorts by printed set, with a precise simulated population.
Explore →CardsWhich cards appear in strong leader lists?
Start from leader evidence, then inspect the card database and individual printings.
Explore →MethodsCan I reproduce and cite this?
Read population rules, confidence intervals, model mixtures and evidence provenance.
Explore →What makes this different
Every number keeps its receipts.
Outcomes come from the game engine, not hand-authored matchup guesses.
Format, match kind, sample floor and filters are visible beside the claim.
Wilson intervals stop tiny or noisy samples from masquerading as certainty.