All guides

Winter Siege — Win Conditions, Battle Strategy and the Matchmaking Problem

Revised after the first battle. The win check is three steps — stars, then power — so equal stars means the stronger alliance wins. One front stronghold opens the HQ, and morale floors at 60%.

공략노트·Source: Ton (TontonYouTube) Day 1 registration and ratios (2026-08-20) + Day 2 battle phase and win conditions (2026-08-22) · cross-checked against the official Winter Siege help centre · provisional, first cycle

🆕 Day-2 update — what Day 1 got wrong (2026-08-22)

After one battle cycle, several core items from the Day-1 write-up turned out to be wrong.

ItemWhat Day 1 said✅ Actual
Win condition"most stars remaining wins" (one tiebreak only)Three-step check — stars → power → who got there first. Equal stars means the stronger alliance wins.
Stronghold unlockmust clear all three front strongholds to reach the HQClearing just one front stronghold opens the HQ
Morale floor35% or below (outside sources)60% is the floor — stat reduction caps at −40%
Star valuessame in every strongholdHQ stars are worth far more (⚠️ provisional)
Attack attemptsTon "20" vs outside "3+7" — contradictory10 per march = 20 per player — both were right

The win condition is the big one. If both sides clear each other's board the score is 0:0, and then the higher-power alliance wins. And in practice clearing a board is easier than expected. The net effect: the weaker alliance loses by default — which is exactly what players are upset about.

🏰

Bottom line — Alliance Championship in fortress form. Three front strongholds plus an HQ, 30 squads each, 2 marches per player → only the top ~60 play. Individually: ① register both marches before the Thursday lock, ② test your ratios first, ③ gear on the right heroes before the lock. As an alliance: judge your odds the moment matchmaking resolves on Tuesday — ahead on power, clear the board; behind, protect stars.

🗓️ Eligibility and schedule lock Thursday, settle Sunday reset

ItemDetail
AlliancesOnly the top 20 by power in the state
GenerationGen 3 and up
FrequencyMonthly, alongside Alliance Showdown week
Slots4 strongholds × 30 squads = 120 · 2 marches each → top ~60 players
Preview / registerMonday – Tuesday
MatchmakingTuesday 23:00–24:00 UTC
PreparationWednesday – Thursday
🔒 LockThursday 24:00 UTC — squads, heroes, troops, bonuses frozen
ScoutingFriday (mission: scout ×2)
BattleSaturday, 24 hours (mission: attack ×10)
Final settlementSaturday → Sunday reset (midnight)

⚠️ Leaving the alliance wipes your registration

Anyone hopping alliances for Alliance Showdown rewards must re-register on return. If the red dot is back on "Go Prepare", you are not registered any more.

🏆 Win condition — a three-step check ★the big Day-2 change★

Poorly explained in-game, and changed by a hotfix on the Monday of the event week.

OrderCriterion
1stMost stars remaining wins
2ndEqual stars → higher power wins
3rdIdentical power → whoever reached that star count first wins

💥 Why this is a problem

  • Both sides clear each other → 0 stars : 0 stars → 2nd check = power → stronger side wins
  • Both sides hold 5 stars → 5 : 5 → stronger side wins again
  • And clearing a board is not hard. Ton's alliance wiped one enemy stronghold in under six hours
  • Conclusion — unless the weaker side can build a star lead, it loses by default
🚩 A "Win" banner during the battle is not final. Ton's alliance saw the win flag the moment they broke the enemy HQ — but they still had 5 stars of their own standing on low morale, and the result could still flip if the enemy cleared them. The check runs again at the Sunday reset.

Is there a better rule? — Ton's take

ProposalProblem
Equal stars → weaker side winsWinning every fight and still losing feels just as unfair from the other chair
Equal stars → whoever cleared firstRewards alliances whose members are awake at reset — a timezone lottery
Ton's proposal: rebalance attack vs defenceMake clearing a board genuinely hard so the star race actually happens

🗺️ Map — how strongholds unlock Day-1 correction

Stronghold 1front · open from the start
Stronghold 2front · open from the start
Stronghold 3front · open from the start
▼ clear any one of them and the HQ opens
Stronghold 4 — HQstars worth 3× · put your best here
  • All three front strongholds are open the moment battle starts.
  • Clearing just one of them opens the HQ. The in-game rules read like sequential unlocking, but that is not how it works.
  • Capacity is 30 squads. From the 31st on, the weakest by power are pushed out as "squad rejected" and never fight.
  • A stronghold with no garrison gets filled by auto-generated squads.

✅ Strategic upshot — pick the weakest stronghold and focus it

Judge it from the average power of its roster, or from how many hard-to-kill players sit in it. Markers/labels let you push an attack order (1→2→3) out to the alliance.

⭐ Stars and ❤️ Hearts

Star values differ by stronghold ⚠️ provisional — needs confirming next cycle

LocationRegular squadWarden
Front strongholds (1–3)⭐ 1⭐ 2
HQ (stronghold 4)⭐ 3⭐ 6

HQ stars are worth . At full capacity a front stronghold is 31 stars; the HQ is in the 90s.

So your very best squads belong in the HQ. This largely settles the Day-1 "concentrate or spread" question — HQ first.

Hearts (lives)

SquadHearts
Regular garrison squad❤️ 3
Warden squad❤️ 5

A squad that loses all hearts is removed — but every heart lost resets morale to 100%. See below.

📉 Morale — Ton's measurements ★Day-1 numbers corrected★

Morale % is stat %. At 90% morale that squad fights with 90% of its stats.
Successful defencesDropMorale
1st−10%p90%
2nd−10%p80%
3rd onward−5%p each75 → 70 → 65 → 60%
Floor60%goes no lower → stat reduction caps at −40%

The "35% floor" I wrote on Day 1 from outside sources was wrong.

⚠️ Morale resets to 100% — you cannot "soften then kill"

Losing one heart snaps morale straight back to 100%. To remove a strong squad you have to repeat the soften → kill cycle once per heart.

Removing one warden (5 hearts) = (2 softening attacks + 1 kill) × 5 ≈ 15 attacks

Ton puts it as "ten attacks to weaken, five to strip the hearts" — one squad can eat the entire attack budget of several top players.

🚩 The unkillable-squad problem

Reported on Ton's Discord — if an enemy squad still beats you at 60% morale (−40% stats), that squad is permanently unkillable. Morale cannot drop further, so no amount of attacking gets through.

One such squad means that stronghold never reaches zero stars. Five or six stars survive, and if your side gets cleared → guaranteed loss. Ton's only description for it: bad matchmaking and bad luck.

🎯 Attack attempts — the numbers, settled

UnitAttempts
Per march10
Per player (2 marches)20
Per alliance (30 players)~600
  • 3 granted at reset, then roughly 1 every 2 hours.
  • Regeneration finishes around the first 14 hoursyou have the remaining 10 hours to spend 10 attacks. Start late and you simply waste them.

Attack supply vs heart supply Ton's Discord log

ItemValue
Alliance attacks available600
Enemy hearts to remove368 (~92 per stronghold)
Slackroughly 40% more attacks than hearts
So 40% of your attacks can miss or be spent purely on morale chip and you still clear. That slack is the numeric reason boards fall easily — and the reason weaker alliances cannot hold stars. Ton reads the balance as tilted too far toward the attacker.

⚔️ Battle strategy — dividing the roles

Pick your scenario first the moment matchmaking resolves

SituationGoalPlay
Ahead on powerFull clearDrive it to 0:0 and take the 2nd check (power) — the most reliable path to a win
Behind on powerProtect starsYou lose the clearing race for certain. Pick off their cheap stars and defend your own
⚠️ Attempting a full clear from behind is the worst option available. You fail to clear, your own board falls, and you lose on stars and on power.

Roles by strength tier

LBottom 30

Morale chip duty. Throw them at targets they cannot beat solo and grind −10/−5%p per hit. Losing is not wasted.

MMid tier

Hunt targets slightly weaker than you and close them out. Reaching upward burns attempts for nothing.

HTop tier

Finish softened heavyweights and hold the HQ stronghold.

Tip: Use markers to publish an attack order per stronghold, and split the roster into numbered teams with assigned targets so attacks don't scatter.

🎲 Outcomes swing wildly

Ton's own measurements, between similarly powered marches:

  • One fight won with 10–15% of his troops left
  • Another lost with the enemy holding 60%+ of theirs

That is a ~75-point spread at the same power level. With this many hero skills interacting, the variance is far higher than Labyran or IWL.

Practical read — a target you "think" you beat is a coin flip. Send attacks at fights you are sure of, and hand the marginal ones to the morale-chip crew.

🏹 Ratios and heroes no event-specific ratio exists

Whatever performs in Labyran, Gaia's Heart, Land of the Brave and IWL performs here.

Gen 3 – 5

MarchRatio (inf/lancer/marks)Heroes
March 150 / 20 / 30your best three
March 250 / 20 / 30next best three
  • Variants: 60/10/30 · 45/10/45 — go with whatever your own testing favours
  • Avoid: blue (rare) heroes and anything without a combat Expedition skill. Epics run one utility + one combat skill and lose out once you have legendaries
  • Favour: Zinman, Alonso, Greg — skills that buff all troop types

Gen 5 and up — marksman ratio + lancer ratio ⚠️ questioned on Day 2

MarchRatioHeroesGear needed
March 1
marksman ratio
48 / 4 / 48best marksman + best infantry + MiaPrimary set on the infantry and marksman heroes
March 2
lancer ratio
60 / 40 / 0best lancer + second infantry + an all-troop-buff marksman heroSecond infantry set + primary lancer set
the marksman hero needs no gear

Why this pairing — the two marches never compete for gear. With zero marksman troops in a lancer ratio there is no reason to gear that hero at all. It squeezes full value from a limited number of maxed sets.

⚠️ Day-2 caveat — a question mark over 60/40

Reviewing a handful of live battle reports, Ton found the 60/40 lancer ratio performing worse than expected and flagged it. His own reports are unusable — he registered his marksman march with his second-string gear by mistake. Needs re-testing next cycle.

Marksman hero priority for the 60/40 march

#HeroWhy
1Bradley (G7)all three skills buff every troop type
2Greg (G3)same — all-troop buffs
3Alonso (G2)same — all-troop buffs
4Zinman (G1)fallback

A "marksman attack +x%" skill does nothing with zero marksman troops on the field.

Whales — 50/20/30 on both marches may be better. Ton sees that style appearing around Gen 15–16; at Gen 11–12 the Mia/Molly marksman ratio and the Bradley lancer ratio still win.

Testing ratios — the free bounty loop

1Pick a bounty

Find a bounty monster out on the map.

2Attack with the test ratio

Send the march you want to measure.

3Stamina refunds

Lose, and the returning march refunds its stamina → loop it for free.

4Average 5–6 runs

As shown above, a single result tells you nothing.

5Cross-check

Gaia's Heart, Land of the Brave, Labyran, IWL. During the event, sparring attacks work too.

🔧 What applies and what doesn't

❌ Does NOT apply

  • Hero exclusive gear (widget) skills — the text at the bottom of the widget card
  • Territory bonuses
  • State position / President skills — titles included
  • City bonuses — gem-bought +10% / +20% attack and friends
  • Frostdragon Tyrant buffs
  • Combat buff items

✅ Does apply

  • Flat widget statshealth and lethality pass through
  • Pet skillsauto-activated
  • Hero skills — the Expedition set
  • Hero gear, charms, chief gear, troop researchfrozen at the end of Preparation

📌 The snapshot is taken at the end of Preparation — Thursday 24:00 UTC

Not the moment you press register. Up to the lock you can keep swapping gear, heroes and troops; after it, nothing you change counts.

Ton himself lost a whole week to gear left on the wrong hero at lock time. Check once at registration, and again on Thursday.

🧭 R4/R5 — building the lanes

  • Each stronghold is auto-sorted by power in Manage Squads
  • Pile 31+ into one and the overflow is benched → it is common to find someone strong sitting out while a weaker squad plays elsewhere. Sweep the list right after registration closes
  • Batch adjust moves several at once — same as Championship lane building
  • Invite from "members not ready", but chase only the people who'd land in the top 60
  • Allocation rule (settled on Day 2): HQ stars are worth 3×, so put your very best in the HQ. But since the enemy only needs one front stronghold to reach it, gutting any single front lane just hands them the doorway
Practical trick — dump everyone into one stronghold first, then deal them out left → right → centre in power-matched groups.

📊 Using the battle reports

  • Each stronghold's battle record shows both attack and defence reports — who fought with what ratio and how it went
  • There are separate sparring attacks — use them for ratio testing
  • ⚠️ Reports appear to be kept for only ~9 hours. The cutoff is by time, not by count → review them live during the battle if you want the data

What to look for: the ratio used (48/4/48 or 60/40) · whether gear was actually equipped · what the morale % was — it changes how you read a win or loss.

⚠️ Known problem — matchmaking imbalance

Ton flagged "some people have been very unfairly matched" on both days.

#Structural cause
The matching key and the deciding factor are different axes. The official rule groups by server progress then pairs on similar Star Rating — but Star Rating is a cumulative record (+1 win / −1 loss), not power. On the first cycle everyone started level, making it effectively random within a group
The 2nd win check is raw power, so the gap becomes the result
One unkillable squad makes a clear impossible
40% more attacks than hearts makes clearing easy → the star race never happens
Byes — an odd group size decides a result with no battle at all
Ton's outlook — Alliance Showdown matchmaking was also terrible at launch; he expects this to settle as Star Ratings accumulate. But it runs monthly, so sorting will take months — and until then it keeps pushing players toward "just join the strongest alliance."

✅ Personal checklist

  • Confirm your alliance is in the state's top 20
  • Check the Tuesday matchmaking result → decide ahead or behind
  • Before the Thursday lock, move gear onto the right hero in each march
  • Register both marches
  • Left and rejoined the alliance? Verify you are still registered
  • Everything finalised before Thursday 24:00 UTC
  • Friday: scout ×2 mission
  • Saturday: start early — regeneration ends in the first 14 hours, so a late start throws attempts away
  • Saturday: spend all 20 attempts (10 per march)
  • Use sparring to confirm your ratio before picking real targets
Source: Ton (TontonYouTube) — Day 1 "registration & ratios" (2026-08-20) + Day 2 "battle phase & win conditions" (2026-08-22) · cross-checked against the official Winter Siege help centre section
Compiled 2026-08-22 · ⚠️ provisional, first cycle (win conditions have already been changed once by hotfix)