App 0.30. New four-faction and custom matches use these rules; earlier saves retain theirs.
The candidate reduced faction spread and opening advantage in this AI sample. Skyrunners still lead. This is evidence about starter decks under heuristic bots, not optimal human balance, human playing time or commercial readiness.
6,144 matches per ruleset; 3,072 appearances per faction. Both rulesets used the same 16 fresh seed blocks (9000–9015), all six matchups, both Companion choices, both seat orders and starting players, and all four scoring/aggressive bot pairings.
| Faction | Before | After |
|---|---|---|
| Skyrunners | 80.2% | 61.7% |
| Veilborn | 42.4% | 48.1% |
| Ironmaw | 10.6% | 45.7% |
| Hearthbound | 66.7% | 44.5% |
First-player wins: 61.1% before; 48.6% after. Mean numbered rounds: 6.34 before; 5.91 after. Round count does not measure human playing time.
| Matchup | First named wins | Second named wins |
|---|---|---|
| Hearthbound vs Skyrunners | 404 / 1024 | 620 / 1024 |
| Hearthbound vs Veilborn | 435 / 1024 | 589 / 1024 |
| Hearthbound vs Ironmaw | 527 / 1024 | 497 / 1024 |
| Skyrunners vs Veilborn | 670 / 1024 | 354 / 1024 |
| Skyrunners vs Ironmaw | 605 / 1024 | 419 / 1024 |
| Veilborn vs Ironmaw | 535 / 1024 | 489 / 1024 |
| Faction | Companion | Wins / games | Rate |
|---|---|---|---|
| Skyrunners | Nima | 955 / 1536 | 62.2% |
| Skyrunners | Vey | 940 / 1536 | 61.2% |
| Veilborn | Nyxara | 742 / 1536 | 48.3% |
| Veilborn | Morrow | 736 / 1536 | 47.9% |
| Ironmaw | Rauk | 683 / 1536 | 44.5% |
| Ironmaw | Skara | 722 / 1536 | 47.0% |
| Hearthbound | Korr | 683 / 1536 | 44.5% |
| Hearthbound | Brindle | 683 / 1536 | 44.5% |
Exploratory selection tested 16 configurations over 12,288 matches using four seed blocks from each of two ranges (2000 and 3000). It compared combat-scoring caps, extra Insight, scout durability, Veilborn costs, opening draw/Charge compensation and first-turn Explore delay. Selection results are not independent confirmation.
The frozen candidate and unchanged rules then each ran 6,144 matches on fresh seeds 9000–9015 with crossed policies. The shipped implementation reproduced the candidate faction win counts and first-player rate exactly in a further 6,144 matches. Repeating seeds verifies implementation, not independence. No further balance changes followed confirmation.
Bots use public board state and their own hand, not opposing hands or future draws. They are heuristic, not expert players. Shared seed blocks and repeated decks create dependence; percentages are descriptive, not confidence guarantees. Custom decks and human choices may change the ranking.
Skyrunners remain strongest. Hearthbound matchups against them and Veilborn need human attention; realm combinations remain under-tested. Avoid further tuning to these confirmation seeds.
Mechanics exercised: realm 342, dodge 45, cleave 858, rally 691. Rules/controller checks cover opening rules, scoring caps, defeat types, serialization and Vanta. Browser visual review and human paired games remain pending.