THE FIELD NOTES · CRITTERFRONT

Hydra and Dragon: a historical late-game mix

A six-slot late-game reference combining two Hydra upgrades, close-range fire and a high-star Hermit King. Its evidence predates the Hermit nerf.

Pre-nerf simulation snapshot · 6–7 September 2026

Every score below predates Hermit King’s 7 September change to 10%/20%/30%/50% defender reflection and 5% less health in both forms. These guides do not claim post-nerf win rates. They also predate the balance update in the 2026-10-07 snapshot: synergy breakpoints moved to 2, 3 and 5 families, trait effects were reworked and many unit stats changed. The battles also ran on the earlier five-row board; the game now has six rows, with a longer route to the Nexus and one more marching row.

This lineup includes Hermit King and needs a retest. Its recorded strength is historical, not a recommendation based on the new values.

The unit reference uses the 2026-10-07 project snapshot. The current playtest build was built 2026-10-14 and may use different rules. Read what was tested ↗

Test budget60g

Net roster investment60g

Base-copy purchase total80g

Round rules testedRound 7

Net investment deducts merge refunds from the purchase total. 0g of the test budget is unspent. This is an owned-roster comparison, not a starting purse or a promise you can find these copies in that round. Shop rolls, reroll costs, liquidity and other players’ purchases were not simulated.

Historical late game · 6 DEPLOYED RECRUITS

The exact roster

Demon

Star III · 4 base copies

Evergrowth / Ravager

Kill-fed defensive attack speed and marching flame trails.

Dragon

Star III · 4 base copies

Ironplate / Ravager

Crowd-control immunity and a one-hex fire aura in both combat forms.

Hermit King

Star IV · 8 base copies

Ironplate / Binder

A reachable taunting shell on defense and fifth-shot reflection on offense.

Hydra

Star II · 2 base copies

Evergrowth / Hatcher

A second recovering marcher and two-target defensive attacks.

Hydra

Star III · 4 base copies

Evergrowth / Hatcher

The larger Hydra investment, with three-target defense and stronger first-death recovery.

Seed

Star III · 4 base copies

Evergrowth / Binder

Vine control on defense and healing sprouts for followers.

Open this roster in the planner ↗

Exact families and stars, with an illustrative starter layout. Edit both formations before playing. This is not a recorded simulation layout. The planner’s base-copy value is the 80g purchase reference, not 60g net investment.

What the families activate

Attacking threats

Ironplate: 2 distinct families · 2-family bonus
Evergrowth: 3 distinct families · 3-family bonus

Defending roles

Binder: 2 distinct families · 2-family bonus
Ravager: 2 distinct families · 2-family bonus

Duplicates do not add faction members. These are current rules interactions, not proof that a particular bonus caused the recorded score.

Give each side a job

Placement notes below are practical interpretation of the abilities, not experimentally optimized hex positions.

YOUR DEFENDING TOWERS

Cover the route

Placement suggestion, not a tested hex layout: cover Hermit King’s engagement area with Hydra attacks and bring Dragon’s one-hex aura close enough to reach passing marchers.
Demon still needs timely kills to offset its self-burn. Do not hide it so far back that it never contributes before losing its health.

YOUR ATTACKING MARCHERS

Plan the approach

Dragon can resist control while applying its short-range fire aura; Hydra provides two original marchers with separate recovery triggers.
The two Hydras count as only one Evergrowth family. Demon, Hydra and Seed reach the three-family regeneration tier, not the five-family bonus.

Recruit toward it, without forcing it

This is a conditional recruitment plan, not a tested shopping sequence. Keep a playable board while the shop decides which copies are available.

Suggested collection priority, not a tested buying sequence: treat this as a late-game target assembled from strong available pieces, not a demand to force every listed copy.
The two Hydras are legal because their stars differ and represent six base copies together. Hermit King adds eight copies, its whole current shared pool; Dragon availability and pool competition remain real constraints.

Where it can struggle

Hermit King now has much less health (160 tower and 145 marcher at star I) and lower defender reflection at stars two through four. The historical score has not been revalidated.
Late-game opponent panels differ because shared-pool conflicts exclude some armies. Indirect scores are not direct finalist head-to-head results.

Recorded matchup notes

Its historical 77.1% points made it the highest-scoring frozen finalist in this stratum, not a globally optimal army.
The audit did not cleanly separate the strongest three late-game finalists on their common opponents. Do not infer a universal late-game winner.
Birdie I + Mushroom I + Petal II + Soul Mage III + Spider III + Wyrm IV held this guide to 33.3% points in eighteen direct holdout battles. That matchup also predates the Hermit King nerf.

What the experiment measured

77.1%

Historical points
A win is 1 point. A draw is ½.

Opposing compositions: 48
Battles in this comparison: 864
Opponent-panel 95% interval: 72.3%–81.9%
Cap or allocation-drop battles: 0

Each opposing composition spends at least 85% of the test budget. Each opposing composition has equal weight. Repeated seeds and swapped seats are not independent opponent strategies. The interval describes this tested panel, not ranked players or a guaranteed result.

Formation sensitivity · the same roster under three placement policies
Test policy	Historical points
Planned defense · front-row offense	72.2%
Random legal defense · randomized Nexus	88.5%
Planned defense · randomized full-depth offense	70.5%

Scores pool many placements. None of these numbers is a result for the editable starter layout linked above. They compare both armies’ applied Nexus damage, not total DPS.

Evidence source and reproducibility

Frozen run: validation-20260906
Stage: validation
Roster: r-118cdfcd7c593eb89250

Download the compact evidence snapshot, including source SHA-256 hashes. Read the method and limitations.