Balance target: approach 50% for every faction and companion. Initially aim for 45–55% overall, review individual matchups outside 35–65%, and investigate companion and starting-order gaps. These are development targets, not achieved balance claims.
11,648 automated matches: all 21 cross-faction pairings, all four companion combinations, both player seats, both starts, four policy pairings and 8 distributed seed blocks. Seven additional same-faction companion pairings are tested separately. Mirror games are excluded from faction averages.
| Faction | Win score | Games | Starts first | Starts second | Scoring policy | Aggressive policy |
|---|---|---|---|---|---|---|
| Hearthbound | 61.0% | 3072 | 65.5% | 56.4% | 60.2% | 61.8% |
| Skyrunners | 56.8% | 3072 | 57.2% | 56.4% | 56.3% | 57.4% |
| Veilborn | 43.0% | 3072 | 46.8% | 39.1% | 45.2% | 40.8% |
| Ironmaw | 69.1% | 3072 | 73.0% | 65.3% | 65.3% | 73.0% |
| Hourglass Covenant | 47.9% | 3072 | 54.9% | 41.0% | 47.5% | 48.3% |
| Revelwild | 25.7% | 3072 | 30.7% | 20.7% | 25.6% | 25.8% |
| Prismforge | 46.5% | 3072 | 54.9% | 38.2% | 44.9% | 48.1% |
Win score counts a win as 1, a draw as ½ and a loss as 0. Each match contributes complementary results for its two players; those are not independent observations.
Read across a row to see its performance against each opponent. An overall average can hide severe matchup weaknesses.
| Faction | Hearthbound | Skyrunners | Veilborn | Ironmaw | Hourglass Covenant | Revelwild | Prismforge |
|---|---|---|---|---|---|---|---|
| Hearthbound | — | 57.4% | 60.4% | 36.3% | 70.3% | 75.8% | 65.6% |
| Skyrunners | 42.6% | — | 61.7% | 34.6% | 64.1% | 78.5% | 59.4% |
| Veilborn | 39.6% | 38.3% | — | 30.5% | 37.9% | 73.2% | 38.3% |
| Ironmaw | 63.7% | 65.4% | 69.5% | — | 59.0% | 78.7% | 78.5% |
| Hourglass Covenant | 29.7% | 35.9% | 62.1% | 41.0% | — | 71.3% | 47.5% |
| Revelwild | 24.2% | 21.5% | 26.8% | 21.3% | 28.7% | — | 31.6% |
| Prismforge | 34.4% | 40.6% | 61.7% | 21.5% | 52.5% | 68.4% | — |
| Faction | Companion | Cross-faction score | Games | Versus own faction’s other companion |
|---|---|---|---|---|
| Hearthbound | Brindle — Companion | 61.2% | 1536 | 49.2% |
| Hearthbound | Korr — Companion | 60.7% | 1536 | 50.8% |
| Skyrunners | Vey — Companion | 57.0% | 1536 | 50.0% |
| Skyrunners | Nima — Companion | 56.6% | 1536 | 50.0% |
| Veilborn | Nyxara | 42.7% | 1536 | 53.1% |
| Veilborn | Morrow | 43.2% | 1536 | 46.9% |
| Ironmaw | Rauk | 66.0% | 1536 | 41.8% |
| Ironmaw | Skara | 72.3% | 1536 | 58.2% |
| Hourglass Covenant | Orryx, the Unfinished | 51.6% | 1536 | 50.0% |
| Hourglass Covenant | Saela, Keeper of Tomorrows | 44.3% | 1536 | 50.0% |
| Revelwild | Zazri, the Grand Misrule | 25.5% | 1536 | 50.0% |
| Revelwild | Pippa, Lanternheart | 25.8% | 1536 | 50.0% |
| Prismforge | Vitra, the Shaping Heart | 48.9% | 1536 | 48.4% |
| Prismforge | Keldrax, Walking Citadel | 44.1% | 1536 | 51.6% |
Review these gaps first; this ranking does not prescribe a buff or nerf without paired candidate testing.
These are the lowest and highest scores across the eight seed blocks, not confidence intervals. A wide range means more independent seed blocks and human games are especially important.
| Faction | Lowest block | Highest block | Companion gap |
|---|---|---|---|
| Hearthbound | 34.9% | 85.9% | 0.5 pp |
| Skyrunners | 38.0% | 91.1% | 0.3 pp |
| Veilborn | 24.5% | 53.6% | 0.5 pp |
| Ironmaw | 42.7% | 83.9% | 6.3 pp |
| Hourglass Covenant | 33.9% | 62.0% | 7.3 pp |
| Revelwild | 12.0% | 39.3% | 0.3 pp |
| Prismforge | 29.2% | 61.2% | 4.8 pp |
Do not add hidden comeback bonuses or change draws to force a 50% result. Equal opportunity should emerge from meaningful decisions and counterplay. A skilled player should still win more often.
Mean numbered round: 6.54. Matches reaching round 12: 202 of 11648. Starting-player win score: 54.5%. Simulations check legal action completion, nonnegative Charge, exact card conservation, valid Construct slots, no unresolved defeats, and JSON save/resume round trips every 25 actions.
Two heuristic policies, fixed full decks, no mulligans, dependent paired seeds; not human win rates or proof of skill. Each game contributes two complementary seat results. Validate changes on fresh held-out seeds and human playtests. The policy does not plan full multi-action combinations. Mirrored games, policy pairings and seed blocks are dependent. These percentages are diagnostics, not predicted human win rates. No gameplay changes are made by this report. Browser layout QA was unavailable.