Doomsday Endgame Recovery: How to Come Back When You Are Below the Bar

Direct answer

If the skull is on you and not on the leader, you are in a recovery window, not a lost game: the Doomsday clock never eliminates you during the 30 second warning and it does not kill you from the troop drain alone, so your only job is to climb back above the bar before the 150 second rot deadline takes your land. Pick exactly one of three branches and commit to it for the whole window. The capture branch takes cheap, low-defense tiles on the open border to buy a few percentage points of usable territory when you still have a usable army and the floor has not decayed yet, and it is the branch that wins when the gap to the bar is small enough that a handful of tiles closes it. The ring branch spends your last gold on SAM launchers so that the 350 warheads of an enemy MIRV land outside your 70-to-150 tile ring and you survive the deadline by being unbreakable rather than by gaining land, which is the branch for when you cannot gain territory but you can still afford the belt of launchers. The standoff branch steps back behind the mutual-nuke standoff of the two big players and lets the bar, not you, do the work, because with the leader immune the leader cannot be eliminated by the clock and the real race is for the second-most-territory seat that the clock is about to open. The single number that decides your branch is your current share of usable tiles relative to the bar, and the single clock that decides your timing is the 30 second warning, the only moment in the window when your army is at full strength. Read those two before you spend a single gold piece; a side that spends on the wrong branch inside the wrong window is a side that rot eliminates with a full army, which is the most common and most avoidable way the Doomsday clock ends a comeback, and it is the failure this entire guide exists to prevent. The reason the three-branch structure is the right shape for the answer is that the Doomsday clock does not give a marked side a single correct move; it gives a side a constraint, the 150 second deadline, and the branch is the tool you bring to that constraint, which means the same lobby can be won by three different plays and the deciding input is the number you are looking at when the skull lands, not the branch you preferred. The 30 second warning is the only moment the decision is free, because after it the drain is spending your army and every second you spend choosing is a second your effective floor is decaying from 40 percent toward 5 percent, so a marked side that enters the warning without a named branch is a marked side that will never have a named branch, the window closes on the choice, and the only thing left to do is watch the rot. That is the whole stakes of the guide in one sentence: the recovery is not hard to execute and it is hard to start, and the two numbers, your share versus the bar and the seconds of floor remaining, are the start.

The rot window: 30 second warning, 5 percent floor, 150 second deadline

Everything in this guide fits inside a 150 second window that the v0.34.24 source defines as three sequential phases, and the recovery is only possible because the phases are ordered rather than simultaneous. The first is the warning, controlled by warnSeconds and set to 30 in Config.ts; for those 30 seconds after you fall below the bar you still hold your full troop cap and the clock has not yet touched your economy, so the warning is the only moment in the whole window when your army is at full strength and nothing has been lost, and it is the moment the whole branch decision must be made rather than during it. The second is the drain, which begins the instant the warning expires and bleeds your troops (and your warship health) down toward a decaying floor; the floor is defined by floorStartPercent at 40 percent of your maximum troops at the start of the drain and decays to drainFloorPercent at 5 percent over floorDecaySeconds of 90 seconds, which means your effective army is at its largest the moment the drain starts and at its smallest roughly ninety seconds later, and the warship takes the same kind of steeper-convex-ramp drain so that a floating core also loses the shield that was hiding it. The third is territory rot, which starts once the floor bottoms out and eats your tiles until the side is eliminated at the rotDeathSeconds deadline of 150 seconds. The v0.34.24 source is explicit that the drain alone never kills, so a side can sit below the bar with a tiny army and still not be eliminated until rot does the work; the reason the game includes rot at all is that without it a stalemate would outlive the final wave and the clock would stop mattering. The comeback window is real but it is a single window, and the v0.34.24 release notes call out the troop-cap change that makes the marked side get a usable comeback army rather than a permanent one, which is why the window is wide enough to recover in and narrow enough that hesitation is fatal. This is the version boundary for the entire guide: the numbers are stable at the v0.34.24 tag and the guide does not claim behavior outside it, so a player reading a faster or older client should treat the 30, 40, 5, 90, and 150 numbers as the v0.34.24 ground truth and not extrapolate from a different preset.

Read the board: are you the marked challenger or the leader

The first and most consequential question in a Doomsday endgame is not which tiles to take but whether the skull is even allowed to sit on you, because the v0.34.24 source implements a leader-immune rule: the side with the most territory is exempt from the clock and never skulls, never drains, and never rots. That single rule flips the entire plan, so before you plan a recovery you must establish your rank on the territory board, and the rank is a three-way distinction that each has a different owner. If you are the side with the most usable tiles, you are not recovering; you are holding the crown, and your problem is the completely different one covered by the crown-position MIRV defense, which is about keeping your lead defensible and the SAM ring you are already building rather than regaining a lost one, and the recovery branches in this guide are the wrong tools for that job. If you are one of the other sides and you are below the bar, you are the marked challenger and this guide is for you, and the bar you are chasing is the next rung of the ladder, which on the FFA normal preset runs 2, 4, 7, 11, 17, 25, and 35 percent. If you are above the current rung but not the leader and the next rung has not been reached yet, you are not marked and the recovery is not live; the clock is only working on you at this exact moment, and you have time to reposition before the skull lands, which is a different and calmer problem than the 150 second scramble this guide describes. The reason the distinction matters so much is that the clock is a one-way race: it can open a seat for the challenger but it can never eliminate the leader, so a challenger who mistakes the leader’s posture for their own and spends on land instead of on survival is a challenger who dies on the deadline while the leader, immune, watches. You can read your rank directly off the board because the four-state Doomsday panel, described in the doomsday clock guide, shows your share against the bar and marks the immune leader, so the “am I marked” question has a one-second answer and the only failure is not asking it; a side that spends a warning window re-reading a board it has already lost the right to is a side that has used its only full-strength moment on a question whose answer was visible the whole time.

The decision framework: three branches for a side below the bar

A marked challenger has exactly three ways back above the bar and the framework is to choose the one that matches your current numbers, then commit to it for the whole window, because switching branches mid-window is the way a recovery becomes a spend with no payoff. The capture branch is available when your drain has not decayed far, your army is still usable, and there is an open border with low-defense tiles: you spend troops on cheap land to buy percentage points, and it wins when the gap between you and the bar is small enough that a few captured tiles close it before the floor decays to 5 percent and your defending army is too thin to hold what you took. The ring branch is available when you cannot gain land but you can still afford SAM launchers, which cost 500,000 gold each: you build a belt of launchers whose combined range, from the 70 tile default up toward the 150 tile asymptote as you upgrade at 45 seconds per upgrade, covers your core, so that the 350 warheads of an enemy MIRV arc land outside the ring and you survive the deadline by being unbreakable; this branch is the one the SAM launcher guide teaches as a defensive posture, repurposed here as a survival posture, and the SAMs fire on the 90 tick cooldown so a well-timed ring intercepts the leak warheads rather than the carrier. The standoff branch is available when you cannot gain land and you cannot afford a full ring, and it is the branch that accepts you will not be the leader this game and instead steps you back behind the mutual-nuke standoff so you live to the last seat the clock opens, and it is the branch that costs nothing in gold but costs time, which is why a faster clock preset makes it the wrong branch. The framework is a condition test, not a preference ladder: the capture branch is named by the point gap and the usable army, the ring branch is named by your gold and the enemy MIRV ladder, and the standoff branch is named by the leader’s immunity and your rank as the second-most-territory side, and none of the three is a default, so a player who cannot name the condition that selects their branch is a player who has not yet decided and will spend the window deciding it. Reading which of the three conditions you satisfy is the whole decision, and the next two sections work each branch out with real numbers so you can see the branch you are in and the number that names it in one move, and the mode-and-map section below shows how the four inputs move the branch from one lobby to another.

Scenario A: 30 percent against a 35 percent bar with 60 seconds of floor left

This scenario assumes a normal-preset FFA at roughly the 35:00 mark, where the bar has reached its 35 percent ceiling and you are at 30 percent of usable tiles with the skull now on you, and it is the most common late endgame because the final rung is the one that keeps the clock live until the very end of a contested game. Your warning has already expired, so the drain is live, and because the floor decays over 90 seconds you have roughly 60 seconds of a still-useful floor before your effective army is down to near the 5 percent minimum, and those 60 seconds are the entire budget for the branch. The gap to the bar is 5 percentage points of usable territory, and 5 points is the size of a capture gap: it is small enough to close with troops and large enough that you must spend the floor to close it, so the capture branch is the only branch that can close a 5 point gap and it is only affordable now, which is exactly the alignment the framework demands. You have a usable army and the open border still has low-defense tiles, so you send troops to convert cheap land and you need to buy exactly 5 points before the floor decays below the army size that can defend those 5 points against the counterstrike. The working assumption is that your rival is the immune leader and is not attacking you, because the leader has nothing to gain from spending troops on you and the clock is already doing that job; that assumption is what makes the capture branch available, and if the leader is instead pushing you the capture branch is dead and you fall to the ring branch or the standoff branch, which is the first branch-change trigger to watch for. The failure state is spending the whole floor on land you cannot hold, which means rot takes it back after the deadline and you are eliminated with the tiles already gone; the counter is to capture only as many points as you can defend with the army that remains after the capture, which is the same test as the final-reserve commitment in the FFA endgame commit guide, applied to a land target instead of a player target. The number that decides Scenario A is the defended-share: the points you can hold after the spend, not the points you can take, and it is a number you can compute before you commit the troops, so a player who commits without computing the defended-share is a player who has already chosen to fail.

Scenario B: the 74-SAM fortress and whether your MIRV leaks through the ring

This scenario assumes a two-sided endgame where the leader has built a stacked core the community calls a fortress, with on the order of 70 to 120 SAM launchers reported in player commentary, and you are the challenger who has a ready MIRV and is deciding whether to break that core before the deadline, which is the ring branch played in reverse, from the attacking side instead of the defending one. The single verified fact that decides it is the SAM interception model in the v0.34.24 source: a MIRV carries 350 warheads, and the SAMs intercept individual warheads rather than the carrier, so the question is not “can I stop the MIRV” but “how many of the 350 warheads land inside your ring,” and the answer is a function of the launcher count and their upgrade level rather than a binary you-or-me call. A SAM at its 90 tick cooldown and 45 second upgrade cadence, with range growing from 70 toward 150, gives you a ring whose effective coverage is a function of how many launchers you have and how upgraded they are; the community’s recurring 100 percent coverage language in the build-strategy commentary is the goal, because every warhead that leaks inside the ring lands on your core and the floor is already decaying under it, so a ring that is 90 percent covered at the moment of the strike is a ring that is letting 35 warheads through and a ring that is 95 percent covered is letting 17 through, and each of those numbers changes whether the core survives the deadline. The cost side is the MIRV ladder from the MIRV price ladder guide: each launch is 25,000,000 gold plus 15,000,000 per prior launch, so a second and third launch on top of an existing fleet are 40,000,000 and 55,000,000 respectively and the cumulative spend to a fourth launch is 190,000,000, which is the number that makes the MIRV a tool you spend once or twice, not a stream you keep opening. The decision is therefore a coverage-versus-cost test: launch the MIRV when the leaked-warhead share is low enough that the core survives it and you can still defend the ring with your decaying floor, and hold the MIRV when the leak is high enough that you would be firing the only tool that can break the fortress into a ring that was built to eat it. The number that decides Scenario B is the leaked-warhead share, and the number that prices the tool is the cumulative MIRV ladder, and the two numbers together are the only two numbers that need to be computed before the launch, which is why the scenario is solvable and not a feel call.

Failure modes and the counter to each

The most common Doomsday-recovery failures are not bad branches, they are right branches run in the wrong window, and each has a specific counter that is a one-number check rather than a new plan, so the counter to every failure is the same discipline applied to a different number. The premature-ring failure is spending the whole floor on SAMs while you are still above the bar, which means you are buying a survival tool you do not need and draining the gold that would have funded the capture branch, and the counter is to hold the ring until the skull is actually on you, because the ring is a deadline tool not a midgame tool and a ring built before the mark is a ring built for a deadline that has not started. The overcapture failure is taking more land than the remaining floor can defend, which rot reclaims after the deadline and converts a lead into an elimination, and the counter is the defended-share test, capturing only as much as the post-spend army holds, which is the number that the capture branch is actually named by. The leader-confusion failure is planning a recovery while you are actually the immune leader, which means you spend on regaining a lead you already have and expose your core while the clock, which could never have touched you, does the work for the challenger, and the counter is the one-second rank read before any spend, the same read that the board section above makes the first question of the whole guide. The standoff-misread failure is choosing the standoff branch because it feels safe, when in fact the bar will not open the second seat until the other challenger dies, so standing off can cost you the very seat you were standing off for, and the counter is to confirm from the territory board that a second seat will actually open before you step back, which is a check on the ladder rung not on your own army. The MIRV-ladder failure is firing a second or third launch into a ring that was sized to eat exactly that many warheads, which the 190,000,000 cumulative fourth-launch number makes a real cost not a hypothetical one, and the counter is the leaked-warhead-share test from Scenario B, the same test that decides whether to fire at all. Every one of these failures is a window error, and the counter to all of them is the same discipline of reading the two numbers, your share versus the bar and the seconds of floor you have left, before the branch is chosen rather than after it has already been spent, because the branch is the answer to a question and a question asked after the answer has been spent is a question that has no answer left.

Mode, map, and speed adjustments

The branch you pick is the same in every Doomsday lobby, but the numbers that name the branch move with the mode, the map, and the clock speed, so a recovery plan that works on one board can fail on another, and the four inputs below are the only four that need to be read before the warning starts. In team modes the ladder is the LEVELS_TEAM schedule at 3/6/10/15/21/28/35 percent rather than the FFA 2/4/7/11/17/25/35 percent, and the bar is applied to a team’s combined territory, which means the gap you must close is measured against your whole team’s share and the capture branch is usually a smaller, more tractable gap because the team is larger and the tiles it controls are more spread out; the ring branch, by contrast, is a team-wide investment and the standoff branch is where you stand behind the alliance rather than behind a single rival, and the leader-immune rule is applied to the team with the most combined territory so the immune unit is a team not a player, which changes who you are standing behind in the standoff branch. The speed preset moves the whole window: on the normal preset the bar reaches 35 percent at 35:00 with 168 second ramps and 54 second pauses, on the slow preset at 45:00, on the fast preset at 25:00, and on the veryfast preset at 15:00, so a faster preset compresses the time you have to choose and spend the branch and makes the standoff branch, which is a wait, more expensive relative to the action branches and cheaper relative to the capture branch, because the capture branch is a spend that is the same size on every preset and the window is shorter on a faster preset, which is why the fast and veryfast presets are the presets where the capture branch wins most often and the standoff branch loses most often. The map decides whether the capture branch is available at all, because it is only available when there is an open border with low-defense tiles; a map that is already fully contested at the bar’s ceiling pushes you toward the ring branch, and a map where your core sits on a defensible island with a Port shifts the ring branch toward the stacked-island posture covered by the midgame island comeback guide, and a map with no islands and no open border at the ceiling is a map where the only branch is the standoff branch and the only question is whether the second seat will open before the deadline. The table below fixes the four inputs that move the branch so that a player can read their lobby and name the branch before the warning starts, and the discipline is to read them before the 30 second warning rather than during it, because the warning is the only moment your army is at full strength and the moment you spend it reading the board is the moment the drain starts against you.

The table is the whole mode-and-map section in one read, and the four rows are the four checks that a player runs before the warning starts, so a player who can name the four inputs and the branch they select is a player who has already made the decision the guide is teaching, and the only thing left is to spend the window on the branch rather than on deciding it.