Methodology
Repeat the world.
Change the deck.
Matched comparisons isolate deck-building decisions: same seed, opponents, seat and pilot—different 99.
Interactive scenario explainer
Configure a comparison.
Change the dimensions to see what a registered experiment means. This teaches the setup; it does not launch a new simulation.
Package comparison
Compare the baseline and challenger on matched seeds in all pods, with Kinnan in all seats, scoring credible attempts through T4.
- Assembly and attempt timing
- Protection available
- Mulligans, mana and actions
- Card exposure and failure reason
Promotion gate
A positive screen must survive more pairs and fresh seeds before the recommended 99 changes.
Expand the details
How the lab earns trust.
01What “better” means+
Earlier credible wins rank higher. Inside the scoring horizon, the lab prioritizes actual deterministic attempts, protection, clean mana, fewer actions, mulligan quality, interaction, resilience and breadth across pods and seats.
02Why matched pairs matter+
The baseline and challenger replay the same randomized world. Pairing prevents a lucky seed, favorable seat or weak opponent draw from quietly becoming a “deck improvement.”
03What the pilot records+
- Mulligan decisions
- Mana and action sequence
- Tutors and cards exposed
- Assembly, attempt and protection
- Natural wins and failure reasons
- Execution and rules errors
04How candidates graduate+
Complete architectures are tested first, coherent packages second and isolated cards last. Small screens reject weak ideas; promising changes graduate to thousands of pairs and fresh holdout seeds.
05What this does not claim+
The simulator does not reproduce table talk, politics, bluffing or every possible human judgment. Absolute percentages describe this registered lab. Paired differences are more useful, and no list is globally optimal outside the stated card pool, pods and objective.
See what those measurements produce.
Explore completed results