Claim the Way now awards at most 1 Renown per turn. Nyxara's base Unseal costs 1 Charge, Passage Unseen costs 0 and Unravel costs 1. Gate Sovereign still costs 1 and Threshold Empress 0. Hexmantle keeps its original -1 Force curse. All creature stats and full deck recipes are retained.
Ironmaw remains favored in the combined tested samples, and Hearthbound is still weak. Four-faction balance is not finished. Automated policies and a small number of seed blocks cannot establish human win probabilities. Old saves retain their rules; start a new match.
Every candidate keeps creature stats, full 30-card deck recipes, companion progression and the published opponent AI unchanged. The Netherway always requires an exhausted return to a normal lane after exploring.
Matched fresh seeds, 24 cross-faction Companion pairings, both seats and starts, all four scoring/aggressive combinations. Each condition also repeated with an offline realm-planning policy. Full canonical decks, no mulligans, normal starting state. Candidate patches are isolated in memory. Model outcomes, not human probabilities. Seeds 61000–61007.
| Faction / measure | Baseline v0.35 | Claim cap 1 | Cap + realm discounts |
|---|---|---|---|
| Hearthbound | 38.9% | 39.5% | 39.6% |
| Skyrunners | 52.8% | 54.4% | 55.1% |
| Veilborn | 35.1% | 35.7% | 33.3% |
| Ironmaw | 73.2% | 70.4% | 71.9% |
| Veilborn vs Ironmaw | 19.9% | 21.9% | 17.2% |
| Starting-player win rate | 47.5% | 46.7% | 45.6% |
| Mean final numbered round | 6.59 | 6.79 | 6.78 |
| Companion Netherway explorations | 0 | 0 | 0 |
| Games | 3072 | 3072 | 3072 |
| Faction / measure | Baseline v0.35 | Claim cap 1 | Cap + realm discounts |
|---|---|---|---|
| Hearthbound | 42.2% | 42.8% | 42.4% |
| Skyrunners | 55.8% | 57.4% | 57.4% |
| Veilborn | 27.5% | 28.8% | 28.0% |
| Ironmaw | 74.5% | 71.0% | 72.3% |
| Veilborn vs Ironmaw | 16.2% | 20.1% | 16.2% |
| Starting-player win rate | 48.4% | 47.9% | 46.8% |
| Mean final numbered round | 6.61 | 6.81 | 6.77 |
| Companion Netherway explorations | 1014 | 1018 | 1035 |
| Games | 3072 | 3072 | 3072 |
Matched fresh seeds, 24 cross-faction Companion pairings, both seats and starts, all four scoring/aggressive combinations. Each condition also repeated with an offline realm-planning policy. Full canonical decks, no mulligans, normal starting state. Candidate patches are isolated in memory. Model outcomes, not human probabilities. Seeds 61000–61003.
| Faction / measure | Cap + realm discounts | Realm package + Hex −2 | Realm package + Unravel 1 |
|---|---|---|---|
| Hearthbound | 34.1% | 34.1% | 32.4% |
| Skyrunners | 53.3% | 53.3% | 49.6% |
| Veilborn | 37.4% | 37.9% | 49.5% |
| Ironmaw | 75.3% | 74.7% | 68.5% |
| Veilborn vs Ironmaw | 18.8% | 20.3% | 39.1% |
| Starting-player win rate | 43.4% | 43.1% | 46.7% |
| Mean final numbered round | 6.88 | 6.88 | 6.87 |
| Companion Netherway explorations | 0 | 0 | 8 |
| Games | 1536 | 1536 | 1536 |
| Faction / measure | Cap + realm discounts | Realm package + Hex −2 | Realm package + Unravel 1 |
|---|---|---|---|
| Hearthbound | 36.8% | 36.8% | 34.2% |
| Skyrunners | 55.6% | 55.6% | 54.6% |
| Veilborn | 32.9% | 33.2% | 42.2% |
| Ironmaw | 74.6% | 74.3% | 69.0% |
| Veilborn vs Ironmaw | 20.7% | 21.5% | 37.5% |
| Starting-player win rate | 44.3% | 44.1% | 46.7% |
| Mean final numbered round | 6.88 | 6.87 | 6.88 |
| Companion Netherway explorations | 569 | 573 | 592 |
| Games | 1536 | 1536 | 1536 |
Matched fresh seeds, 24 cross-faction Companion pairings, both seats and starts, all four scoring/aggressive combinations. Each condition also repeated with an offline realm-planning policy. Full canonical decks, no mulligans, normal starting state. Candidate patches are isolated in memory. Model outcomes, not human probabilities. Seeds 71000–71003.
| Faction / measure | Baseline v0.35 | Realm package + Unravel 1 | Cap + Unravel 1 |
|---|---|---|---|
| Hearthbound | 34.9% | 36.7% | 36.7% |
| Skyrunners | 48.2% | 52.3% | 50.3% |
| Veilborn | 51.6% | 54.7% | 57.3% |
| Ironmaw | 65.4% | 56.2% | 55.7% |
| Veilborn vs Ironmaw | 34.4% | 42.2% | 43.8% |
| Starting-player win rate | 47.4% | 43.1% | 44.4% |
| Mean final numbered round | 6.52 | 6.64 | 6.64 |
| Companion Netherway explorations | 0 | 0 | 0 |
| Games | 1536 | 1536 | 1536 |
| Faction / measure | Baseline v0.35 | Realm package + Unravel 1 | Cap + Unravel 1 |
|---|---|---|---|
| Hearthbound | 38.7% | 40.6% | 40.5% |
| Skyrunners | 51.8% | 56.2% | 56.2% |
| Veilborn | 42.8% | 45.4% | 44.5% |
| Ironmaw | 66.7% | 57.7% | 58.7% |
| Veilborn vs Ironmaw | 30.5% | 37.9% | 34.8% |
| Starting-player win rate | 50.3% | 47.7% | 50.0% |
| Mean final numbered round | 6.52 | 6.59 | 6.64 |
| Companion Netherway explorations | 529 | 560 | 531 |
| Games | 1536 | 1536 | 1536 |
Claim cap: first own-turn Challenge defeating an opposing Ally gives 1 Renown, maximum 1 per turn. Realm discounts: base Nyxara Unseal 2 → 1 Charge; Passage Unseen 1 → 0. Hex candidate: Hexmantle entry curse −1 → −2 Force, strongest only and original expiry. Unravel candidate: 2 → 1 Charge, still Allies only. The two disruption candidates were tested separately.
33792 matches across all experiments. Full cross-faction Companion schedules, both seat orders, both starts, all four scoring/aggressive policy combinations. Rates count a draw as half a win. Mirrored games and shared seed blocks are dependent; these are exploratory results, not independent human win probabilities. The fresh-seed comparison was specified after screening; it did not reuse the screening seeds.
The realm-planning wrapper is only a sensitivity check and is not the published opponent AI. It is one attempted strategy, not optimal play. Hearthbound remains weaker in several conditions; this update does not establish four-faction balance. Old saved games keep their stored card costs and rule edition. Start a new match for this update.
All simulations required legal actions, nonnegative Charge, exact 30-card conservation at completion and a bounded action count. Matched schedules and record counts are verified by the report builder. Targeted rule tests cover cost payment, defender blocking, exhausted return and save compatibility. Browser layout was not tested in this environment.
Main trial records · Disruption screening records · Fresh-seed records · Historical Companion tournament