WHITEOUT SURVIVAL ・ Strategy proven with numbers
In our last SVS, Sunfire Castle occupation time ended 2:00 vs 2:18. We had the strongest rally leader and still lost by 18 minutes. This article reviews that fight phase by phase, in numbers.
In SVS (a battle between two states), the Sunfire Castle fight is decided by total time holding the castle. Last time we held it for 2 hours 00 minutes and they held it for 2 hours 18 minutes. We lost by 18 minutes.
We had one rally leader stronger than anyone on their side (called A below). They were weaker than A but had more players, and kept landing "10-rally chains" (10 rallies in a row) to the very end. The castle changed hands again and again, and we never caught up.
Occupation time: 9 minutes short of a tie
258 minutes in total. Moving 9 minutes from their hold to ours would have tied it. Time "moves" from one side to the other, so an 18-minute gap comes down to a fight over 9 minutes.
We assumed one standout player would be enough to hold the castle. It wasn't. This is for states that rely on one strong player the same way: where we lost time, and what to change next time.
| Term | Meaning |
|---|---|
| Rally | Several players' troops combined into one attack on the castle. Takes at least 5 minutes from launch to landing |
| Garrison | Stationing troops in the castle to defend it. Takes march time only |
| 10-rally chain, double rally | Landing 10 or 2 rallies in a row |
| Ghost rally | A rally the opponent can't see. Explained in 02 |
| Initiative | Held by the side garrisoning the castle. The other side has to launch a rally, which takes at least 5 minutes to land |
| Handoff | Passing a castle taken by another alliance back to your own alliance |
We had three rally leaders. They had two, plus nine other players. At most, they could field a 10-rally chain plus one staggered rally: 11 in total.
| Side | Rally leader | Profile |
|---|---|---|
| Us | A | By far the strongest. With A garrisoned, none of their single rallies could break through |
| Us | B, C | Same alliance as A. They moved to another alliance only to launch ghost rallies. B alone might not break through, so B and C had to go as a double rally |
| Them | D, E | Their two main players. E's solo garrison could not hold against the B+C double rally. Even with A garrisoned, it was borderline |
| Them | Nine others | Joined ghost rallies and multi-rally chains. Enough players to send a 10–11 rally chain at the 1.3M scale against our A |
Power ranking: A, D, E, B, C, then the nine others.
The rules that matter here are below. The Sunfire Castle rules themselves are covered in "Sunfire Castle: Maximizing Occupation Time, Derived From the Rules."
The fight went through six phases. Each time we changed our approach, they adapted within one or two cycles. Phases 5 and 6 ran in parallel near the end.
We took the castle first and A garrisoned. The initiative was ours. With A garrisoned, none of their players could break through alone, and we held the garrison the whole time until their 10-rally chain took it. In the meantime they sent two 4-rally chains. Looking back, these seem to have been staggered rallies aimed at our ghost rally. Partway through, they apparently realized 4-rally chains couldn't take it and landed a full 10-rally chain, likely judging that we had no ghost rally.
After the 10-rally chain took it, we launched A's rally. This is where we should also have launched staggered rallies to answer their ghost rally, but we didn't see it coming. Each time A broke through, their 10-rally ghost chain landed right away and took it back, and E garrisoned again. This loop cost us the initiative and put us behind.
This was the move missing in phase 2: cut their ghost rally with staggered rallies, win a 5-minute garrison, and take the initiative back. We couldn't see their ghost rally, but we could predict that it would land right after A took the castle. So after A launched a rally, B and C sent a double rally 1 minute apart, timed to land after their ghost rally. Even if their ghost rally took the castle, this double rally would take it straight back. They would then have to launch a new rally, which takes at least 5 minutes to land. Those 5 minutes are ours to garrison, and the initiative is ours. B and C stayed in A's alliance, and once they broke the castle, A would garrison immediately. But they timed their ghost rally to land after our double rally.
A ghost rally is normally timed to land right after the opponent takes the castle, with almost no gap. But that is pointless if our staggered rally retakes it right away, so they landed theirs after our double rally. That means the longer the gap between A taking the castle and the double rally landing, the longer we held it. We spaced them 2.5 minutes apart, aiming to hold at least half of the 5-minute prep. They responded by timing the ghost rally to A instead of the double rally, and timing D's garrison to the double rally. We went back and forth several times, changing the gap to 2 minutes, then 1.5 minutes.
After A broke the castle, the moment they formed a 10-rally chain, A pulled out of the garrison and moved to the ghost side. Once a ghost rally broke the castle, we handed it over and A came back to garrison. With A garrisoned, their only option was to form a 10-rally chain to break A, which gave us the turn to send a ghost rally. This pattern started working near the end, but we ran out of time.
B and C moved to another alliance only to send ghost rallies. After a ghost rally broke the castle, they had to hand it to A's alliance. They had timed one staggered rally from E after that, and it landed less than a minute later. Unless the other members entered at the same moment as A, E broke through even with A there. The handoff failed 3–4 times, and each time we had to start over: launch A's rally and retake it with staggered rallies. Because these failures were frequent when B and C sent the ghost rally, we also had A send the ghost rally directly (phase 5), at the cost of more load on A (detailed in 05).
A rally takes at least 5 minutes from launch to landing. For those 5 minutes, the side garrisoning the castle can see the rally and its landing time. A garrison, on the other hand, only takes march time. You can see the opponent's rally and still get a garrison in before it lands. The exception is the ghost rally: the side that isn't garrisoning can't see it, so it can't respond until it lands. That's why the side that can see the other's rally before deciding its own move has the edge.
Time from deciding to move until it takes effect: rally vs garrison
Lengths are approximate. You can see an incoming rally and still send a garrison in time. The reverse doesn't work: launching a rally after seeing a garrison takes 5 minutes.
In this fight, this mechanic worked in their favor at some points and in ours at others.
Our staggered double rally targeted a castle they were garrisoning, so they could see it. Their ghost rally, on the other hand, was invisible to us. However we changed the gap, they could see it and time their ghost rally or D's garrison to match. We could only set our gap by guessing.
When only our moves are visible, they get to respond last. They had many moves: the 10-rally chain, E's single rally, D's garrison. We had only one set.
In the late-game pattern, less than a minute passed between the ghost rally breaking the castle and E's single rally landing. Whether we could hand the castle back to our own alliance in that window decided each cycle.
The ghost rally came from either B and C or A, and that decided who received the castle.
| Ghost rally sent by | Castle received by | Notes |
|---|---|---|
| B and C (temporarily in another alliance) | A's alliance (A and members) | Handoffs failed often |
| A (moving to another alliance) | B or C, who stayed in our alliance | Used because B and C's handoffs failed often. Puts more load on A |
One cycle (when A sends the ghost rally)
Shaded steps 4–5: under 1 minute from the ghost landing to E's staggered rally landing
If the handoff doesn't finish in time, E breaks through.
Their only option to break A is to form another 10-rally chain. It becomes our turn to send a ghost rally.
We go back to launching A's rally and retaking it with staggered rallies. Retaking takes at least the 5-minute prep plus march time.
An 18-minute gap is small enough that two failed handoffs would have flipped it.
After a failure, retaking takes at least the 5-minute prep plus march time, and they hold the castle the whole time. About 5 minutes that should have been ours move to them, so each failure widens the gap by about 10 minutes (estimate).
How many successes would have won it (occupation time gap, us − them)
The gap by how many of the 3–4 failed handoffs had succeeded, at about 10 minutes per failure.
This calculation has three sources of uncertainty.
Even with the same strength, switching patterns and cleaner handoffs left room to win.
Once they've formed a 10-rally chain, the castle is going to fall anyway. A gains nothing by staying. A should have garrisoned only until the chain was drawn out, then moved to ghost rallies. The pattern that started working near the end was exactly this. Switching earlier would have won back those cycles.
In a window under one minute, any reaction delay becomes a failure.
What helps more than tactics is having a third attacking option (estimate).
Moves available per cycle
B alone might not break through, and C most likely couldn't break E. So we had no backup for a failed handoff.
If B could break the castle alone right after their 10-rally chain, C could be held in reserve to land right after E. A failed handoff would no longer mean starting over. Depth at this level, more than making the one standout player even stronger, translates directly into occupation time.
What our state lost from this defeat, and what it kept, is covered in "What Happens When You Lose SvS."
The phases and failure counts in this article are based on the author's memory and notes. Items marked "(estimate)" are unverified estimates.