Why the Attacker Regrows Faster: Troop Recovery After Contact

Direct answer. The attacker regrows faster because it spends its bar into the flat 30-55% plateau and lets it recover on the steep left side of the regen curve, while a defender who dumps to near-zero collapses the base of its own recovery formula. Your per-tick refill is (10 + troops^0.73 / 4) × (1 − troops / max): it peaks at about 42% of your cap and then tapers to zero at the cap. A bar parked near full or near empty both refill slowly; a bar cycled through the middle refills fast. The v0.34.24 fix also removed an old exploit where spamming an attack at a target you did not actually border pinned your regen at peak and refunded troops past the cap, so the gap you see now is pacing and cap size, not a hidden bonus for attackers.

Why the Regrow Feels Asymmetric in the First Place

The single most common reason one player’s bar climbs faster than yours is not a bonus on the attacker’s side — it is a handicap on the defender’s side, and it is self-inflicted. The engine adds troops to every living player every tick using one shared function, and that function has exactly two pieces: a base term and a taper term. The base term, 10 + troops^0.73 / 4, grows as your bar fills: the more troops you hold, the larger the number the engine is about to add. The 0.73 exponent makes it grow sublinearly, so the marginal gain of each extra troop shrinks, but the direction is unambiguous — a bar with more troops in it has a larger base than a bar with fewer. Run the two terms at two concrete fills against a 100,000-troop cap to see it: at 20,000 troops the base term alone is about 10 + 20000^0.73 / 4 ≈ 355, while at 60,000 troops the base is about 10 + 60000^0.73 / 4 ≈ 779. The bar holding three times the troops carries more than double the base. The taper term, (1 − troops / max), does the opposite: the closer you are to your cap, the smaller it gets, and it reaches exactly zero at the cap. At 20,000 troops out of 100,000 the taper is 0.8; at 60,000 it is 0.4; at 90,000 it is only 0.1.

Multiply those two together and you have the whole behavior, and the numbers below are the exact per-tick values the engine would add at each fill on a 100,000-troop cap with no bot or Nation multiplier applied. At a low fill the base is small but the taper is nearly one, so the product is modest: at 20,000 troops that is roughly 355 × 0.8 ≈ 284 troops per tick. As you fill into the middle, the base grows faster than the taper shrinks, and the product climbs to its peak of about 350 at roughly 42% of the cap. Past the peak, the taper keeps shrinking while the base keeps growing only slowly, so the product declines: at 60,000 it is about 312, at 80,000 about 192, at 90,000 about 104, and at a full bar exactly zero. The practical consequence is that the middle of the bar is where the engine hands you the most troops per tick, and the two edges are where it hands you the least. If your neighbor is constantly cycling their bar through the middle and you are constantly dumping it to the bottom before recovering, you will always watch them refill faster, because the curve simply pays them better. This is why the asymmetry reads as “the attacker is faster”: the attacker is the one who just spent into the plateau, and the defender is the one sitting at the floor or the ceiling. It is not a perception problem; it is the literal arithmetic of the formula, and the rest of this guide is about how to keep your own bar on the better half of that curve so the asymmetry stops being one-sided, because the same function that looks like it favors the attacker is actually rewarding whichever side chooses to stay in the middle.

The Numbers: Where the Peak Actually Sits and Why the Band Is Flat

The peak of troops^0.73 × (1 − troops / max) falls at troops / max = 0.73 / 1.73 ≈ 0.4219, so the theoretical maximum-rate fill is about 42.2% of your cap. That is the number the community keeps rounding to “40%” or “a little under half,” and it is the point of the whole pacing game: hold the bar here and you are refilling as fast as the engine will allow. What is more useful than the exact peak, though, is the shape around it, because the curve is flat across the middle and steep at both ends. For a human with no cities, the per-tick value relative to that 350-per-tick peak looks like this on a 100,000-troop cap:

Fill %	Per-tick troops	Relative to peak	Band
20%	~284	81%	recovering
30%	~332	95%	plateau
40%	~349	100% (peak)	plateau
50%	~342	98%	plateau
55%	~336	96%	plateau
60%	~312	89%	falling off
70%	~261	75%	tapering
80%	~192	55%	tapering
90%	~104	30%	near floor
100%	0	0%	cap reached

Two things stand out, and both of them change how you should time a push. First, between roughly 30% and 55% you are all within about five percent of the peak, so small timing errors inside that band cost you almost nothing: you can push at 35%, hold at 45%, or re-engage at 52% and be doing the same regrow work, because the engine cannot tell the difference well enough for it to matter. That flatness is why the game “feels” forgiving around 40%: you are operating on a wide, near-flat plateau rather than a razor’s edge, and a few ticks of uncertainty about exactly when to commit are absorbed by the curve. The band is not symmetric, though — past 55% it starts to fall away, and by 60% you are down to about 89% of the peak. The second, sharper fact is that the collapse is abrupt at the very top: the drop from 70% to 90% goes from three-quarters of the peak to barely a third, and the final tick into the cap adds exactly nothing. Your worst timing error is not pushing slightly early — it is hoarding the bar to 95% or 100% and discovering the next wave is far away while a neighbor cycles through the middle. The 30-55% plateau is the working zone for this guide: keep the bar in it and you are, for all practical pacing purposes, at the engine’s best rate. If you want a single rule derived from this table, it is that the cost of being a little off-peak in the middle is a rounding error, while the cost of drifting to either edge is a real, compounding loss of troops every single tick, and that loss is the entire budget the rest of this guide is about.

Cap Size Sets Your Absolute Regrow Speed

The same fill fraction is not the same recovery speed on every map, because the cap is a function of territory and of Cities. The human cap is 2 × (tiles^0.6 × 1000 + 50000) plus 250,000 for each finished City level, with bots running at one third of their cap. Work a few concrete cases with no Cities so the scaling is not abstract: at 1 tile the cap is about 102,000; at 100 tiles it is about 132,000; at 1,000 tiles it is about 226,000; at 4,000 tiles it is about 390,000; and at 10,000 tiles it is about 602,000. The 0.6 power on the tile count means the tile contribution grows sublinearly — doubling your tiles does not double your cap — so most of the leverage in the early game comes from the flat 50,000 base term, and the real step change comes from Cities: each finished City level adds a flat 250,000, which at a mid-size core is larger than the entire tile contribution. That is why a single early City shifts the whole regrow calculation more than any amount of careful tile farming around it.

This is the quiet source of the “they just have more troops than me” feeling, and it is the reason a player who took a couple of extra Cities early regrows visibly faster than one who did not. When you are the larger of two neighbors, your per-tick absolute gain while regrowing is bigger, so holding the bar a little lower before a decisive push costs you more — you should weight the regrow window more heavily and re-engage sooner rather than letting a larger bar sit idle, because each idle tick is a bigger number of troops that your bigger cap lets you buy back. When you are the smaller neighbor, your per-tick absolute gain is smaller, so the “wait for the bar to regrow” step is worth less in absolute troops and you can commit the next available wave sooner without giving up as much. The percentage rule stays the same in both cases; what changes with size is how much absolute value is at stake on either side of the push-and-hold decision, and how much a City level shifts the whole calculation. If both of you have Cities, the one with the higher level has a larger cap and therefore a larger absolute regrow, which is why an early City is a pacing advantage as much as a troop-number advantage: it does not just give you more troops to spend, it lets every tick of recovery you choose to wait hand you a bigger number back. The population growth guide owns the cap formula and the City bonus in full; this page uses that cap only to explain why the same 42% feels different on a small core versus a large one, and why “same percentage, different speed” is the correct mental model for two neighbors of unequal size rather than a sign that either of you has a hidden bonus or a hidden penalty on the regen rate itself.

Scenario A: The Two-Sided Pressure Window

You hold 4,000 tiles in an FFA mid-game with a cap of about 390,000, and two neighbors are expanding toward you from the east and the west. Your bar is at 46%, roughly 180,000 troops, which is just past the 42% peak, so your per-tick gain is essentially at the plateau maximum — you are refilling as fast as the engine will let you right now, at about 99% of your theoretical best, and each additional tick of holding adds a nearly full tick of troops. The eastern neighbor shows a bar at 70% and is visibly pressing; the western neighbor sits at 30% and looks like it is consolidating its core rather than committing. The decision is whether to launch a preemptive push now into the western soft flank, or to let your bar regrow for a few ticks and hit harder later. The arithmetic that makes this a real choice rather than a guess is the difference between holding and spending at exactly these two fills: at 46% you are on the plateau, so a few more ticks of holding buy you a near-full tick of troops per second, but you are also one push away from the floor if you spend hard, and that is the risk of the “wait a bit longer” option when a soft target is already in front of you.

Assumptions: no nukes on the field, no Cities under construction, both neighbors human, and the western neighbor will not launch into you within the next few ticks. If you push west at 46%, you commit roughly 45-50% of the bar, landing the western flank low (into the steep left side) where it recovers fast, and you spend the east pressure window regrowing in the middle. The eastern neighbor at 70% is in the tapering zone, so its per-tick absolute gain is actually lower than yours at 46% even though its bar looks fuller — at 70% it is adding about 75% of its peak while you at 46% are adding about 99% of yours, and on a similar cap that means you are out-regrowing the player with the fuller-looking bar. That is the counter-intuitive read: the player with the fuller bar is not always the one regrowing faster, and that is exactly the mistake that makes people think the opponent has a bonus. The push into the 30% western flank is the correct move here, because you are at the peak of your own curve and the soft target is in its own low-regrow recovery band. If instead you hold and wait for your bar to climb to 80%, you slide into the taper, your per-tick gain drops well below the western neighbor’s recovering bar, and the eastern neighbor’s slow-but-still-advancing 70% bar reaches your border first. The fallback branch matters too: if you suspect the east neighbor will strike first rather than continue pressing, the same logic holds — you are already at your peak, so the regrow you gain by waiting is real, but only if the east does not close the distance; the moment the soft western flank is gone, the “wait” option is worth nothing. The rule: at a peak-position bar with a low-fill soft target available, push now, because the engine is paying you the best rate exactly now, not a few ticks from now, and a recoverable flank in front of you is the cheapest push you will get all day.

Scenario B: The Endgame Overextension Trap

Late in a ranked 2v2, your team holds a broad front and your bar has been pushed to 95% because you have been absorbing small attacks instead of committing. Your cap is 200,000, so at 95% you are sitting on about 190,000 troops and the taper is down to 0.05 — the engine is adding only five percent of the base term per tick, which on this cap works out to roughly a couple hundred troops per tick, the slowest part of the whole curve and nearly nothing in the context of a 200,000-troop front. You are tempted to dump it all into a single decisive push against the larger enemy line. The trap: the push may win the line, but it leaves your entire front with a bar that regenerates almost nothing for the next long stretch, which is exactly when the enemy — having banked troops into a mid-range reserve around 50% — can re-engage at its own peak and roll the now-thin line. Concretely, if the enemy is at 50% it is within a fraction of a percent of its peak and every tick it waits it refills at full rate, while your post-push bar at the floor refills at a trickle; the gap between your recovery and theirs widens every single tick, so a won skirmish becomes a lost front even though the immediate fight went your way.

The fix is the opposite of the instinct: before a decisive endgame push, deliberately let the bar sit a few ticks below the 55% mark, or commit the push in two waves with a short regrow window between them so each wave leaves the front inside the 30-55% plateau rather than at the floor. If you commit wave one at 95% and leave the bar at, say, 30%, that 30% bar is back on the plateau and refilling fast, so wave two a few ticks later is stronger than the full dump would have been, because you spent the first wave from a full bar and recovered the second wave from the middle. The winning overtime guide covers the endgame tempo this scenario sits inside; this page contributes the specific rule that a bar at 95% is the worst place from which to launch the biggest push you have, because that is where your recovery per tick is closest to zero and the enemy’s is closest to its maximum, and the two facts combine into a situation where winning the exchange and losing the war happen in the same move. The same logic applies to any long fight, not just overtime: the player who keeps cycling through the middle holds the tempo, and the player who hoards to the ceiling and then dumps to the floor hands the tempo to the opponent. In a 2v2 you can make this explicit with your partner — agree that neither of you dumps below 30% on the shared front, so the front’s combined regrow never collapses while the enemy line banks into its own peak — and the discipline turns a fragile one-shot into a repeatable rhythm that the enemy cannot answer by banking, because every time they bank, your mid-plateau bar is already recovering and ready to commit again on your schedule, not theirs.

Failure Mode and Its Counter

The failure that produces the “they always regrow faster than me” complaint is dumping to near-zero. The moment your bar approaches zero, the base term troops^0.73 / 4 collapses with it, and you are left adding almost nothing per tick. A concrete shape of the same mistake: early on you hold 2,500 troops and gain roughly a couple hundred per tick, but if you immediately attack with everything you drop to a few hundred troops and the per-tick gain falls to barely a trickle, so the bar that felt large moments ago now recovers at a crawl while the player who sent a measured 30% and cycled back into the middle is already refilling at full rate. The asymmetry is created in that exact exchange: you are at the floor adding almost nothing a tick, and the other player is at the middle adding a full tick’s worth, so over the next twenty ticks the gap grows by thousands of troops with no new attack from either side. That is the “they just have more” feeling, produced by your own pacing, and it compounds because the player at the floor also has a smaller base, so the next time they try to recover they are again starting from a worse place than the player who stayed in the middle.

The counter is pacing: send in the 30-40% range, send only when your bar has recovered into the plateau, and never spend the bar all the way down before the next tick has had a chance to pay you back. The related but distinct failure is holding a full bar across a contact: you arrive at the fight at 95% or higher and therefore recover the slowest after it, turning a win into a slow bleed, because the bar you won with is now the bar that refills last. Both failures are the same error viewed from opposite edges — leaving the 30-55% plateau — and the counter is the same discipline: keep the bar cycled through the middle, and let the two edges be brief transit states, not resting states. The troop bar attack timing guide is the per-tick timing reference this failure section defers to; here the point is the asymmetry that pacing creates between you and the player who does not pace, and it is an asymmetry you can feel in the first few seconds of a fight. A useful self-check after every exchange is to ask where your bar landed: if it landed in the 30-55% band you paid for the push with almost nothing in future regrow, and if it landed at the floor or the ceiling you paid the maximum. Train the habit of reading the bar’s fill before you commit, not after, and the “attacker is faster” problem largely disappears, because you stop being the defender at the floor and start being the attacker in the middle — the same role you used to accuse your opponent of secretly having, and one that, once you take it yourself, reveals that there was never any hidden bonus, only a habit of staying on the better half of the curve that anyone can pick up the moment they start watching their own bar.

What the v0.34.24 Refund Fix Removed

Before v0.34.24, the attack code held an attack with nothing left to conquer until the next tick instead of canceling it. The shipped comment at the v0.34.24 tag says plainly what that let a player do: held until the next tick, the troops would sit out the regen step and come back free, so one such attack per tick kept regen at its peak rate and refunded past the cap. In other words, spamming the attack slider at a target you did not actually border — or at a target you had already surrounded so that the conquest queue went empty — kept your regen pinned at its peak and let troops accrue past the cap. That was a genuine, measurable regrow advantage for the player doing it, and it is one of the real reasons an attacker’s bar could look like it was refilling faster than a careful defender’s: the defender was pacing correctly, but the attacker was also farming a free refill on top of good pacing, so the gap was not just skill, it was a bug that the defender could not see from their side of the map.

v0.34.24 closed it: AttackExecution now retreats immediately the moment the conquest queue is empty, so the held-until-next-tick trick no longer works and the refund is gone. The boundary that matters for this guide is precise: the pacing and cap-size asymmetry described above is stable across these versions, but the refund exploit is v0.34.24-specific. If you are replaying an older match where an opponent’s bar seemed to regrow impossibly fast against a surrounded target, that is the exploit, not a pacing difference, and no amount of your own pacing would have closed the gap. On current v0.34.24 and later, an empty conquest queue means an immediate retreat, and a retreat forfeits 25% of the attacking troops, so spamming attacks is now a net cost, not a regrow trick: every bogus attack you fire not only fails to conquer, it also burns a quarter of the troops you committed. That changes the read of an aggressive opponent — if they are spamming the slider with no real front, they are now bleeding 25% per wasted attack, which is a sign of desperation or a misclick, not a hidden regrow source. The practical rule for you is to stop treating “they must be regrowing faster” as the explanation for an opponent’s numbers: on current versions the only regrow inputs are pacing (where on the curve) and cap (how big), and the refund lever is gone. Recheck the attack-construction path after any release that touches it, because that is where the fix lives, and if a future release reintroduces a hold-on-empty path the refund advantage comes back with it — but on the versions you are playing right now, the refund is not a factor, and any regrow advantage you see is the pacing and cap math this guide has already laid out, which means it is fully readable and fully matchable on your own side of the board.

Mode and Map Adjustments

The plateau rule is mode-independent, but the numbers that make it matter shift, so the same 30-55% guidance lands differently depending on where and how you are playing. In FFA the relevant variable is your cap relative to your two closest neighbors, because every neighbor is a potential target and the soft flank is always available; keep the bar in the 30-55% plateau and push into whichever neighbor sits in its own low-regrow recovery band. FFA also has no teammate to coordinate a shared front with, so the “do not dump the shared front” discipline from the 2v2 case does not apply — instead, your entire bar is yours to cycle, and the failure mode is dumping your only bar to the floor in a botched preemptive strike with no backup coming. The rule for FFA is to keep at least one wave of regrow headroom: never let the bar go below the 30% band unless you have a specific reason, because with no partner to cover the front, a floored bar is an unrecoverable window where every neighbor’s regrow is a threat and your own is at its worst, which is the worst possible combination to be caught in when three neighbors are all watching the same open space you just left.

In ranked 2v2 the variable is team-wide pacing: the front that dumps to zero first is the front that bleeds, so coordinate with your partner to commit in waves that leave both bars in the plateau rather than one side dumping all at once. The concrete version of that rule is to agree on a shared floor — say 30% — below which neither of you commits to the shared front, so the front’s combined regrow never collapses while the enemy line banks into its own peak. On maps with lots of early Cities the cap grows faster, so the absolute per-tick regrow is larger and the “wait a few ticks” step is worth more — hold the bar a touch lower before big pushes, because each tick of holding is buying more absolute troops than on a sparse map, and the City-driven cap bump means your plateau sits at a higher absolute number than your opponents who did not bank Cities. On maps with sparse Cities and slow territory gain the cap is small and the regrow window is short, so you can re-engage sooner and hold the bar a touch higher, because the absolute value of a held tick is lower and the enemy is not scaling its cap the same way you are. The ranked 2v2 guide and the winning overtime guide own the tempo these adjustments feed into; this page contributes only the regrow-speed piece of that tempo, which is the part most players feel but cannot quantify, and the part that, once quantified, is the difference between a front that holds and a front that bleeds, because a front that holds is a front whose combined bar is cycling through the middle on schedule while the enemy front is forced to watch their own banking and their own floor both at the same time.

Putting It Together: A Decision Checklist

Before every push, run four checks, and the order matters because each one gates the next. First, where is your bar? If it is in the 30-55% plateau you are at the engine’s best rate and the push-and-regrow cycle is cheap — the tick you give up to commit is the tick the curve pays you back at near-peak, so the push is cheap and the recovery after it is fast. If it is above 70% or below 20% you are on a slow edge and you should either recover first or accept a slower refill afterward; at 95% the refill is nearly free to give up but nearly impossible to get back, and at the floor the refill is the worst of the whole curve, so those two fills are the ones to avoid committing from, because committing from them means spending from a bar that is already at the slowest part of its own recovery. Second, where is the target’s bar? A soft flank in its own low-fill recovery band is a cheap push now, because you are hitting a bar that is already refilling fast and will not out-regrow you in the window; a target sitting at a full bar is one you should let it spend down first, because hitting a full bar means fighting a line that just committed and is at the slowest part of its own recovery, which is the opposite of what you want in a long fight.

Third, what is the cap difference? If you are the larger neighbor, weight the regrow window more heavily because your absolute per-tick gain is bigger, so each held tick is worth more and you should re-engage sooner; if you are the smaller one, re-engage sooner because your absolute gain is smaller and a held tick is worth less in absolute troops. Fourth, is the conquest queue real? On v0.34.24 and later, an attack with nothing to conquer retreats immediately and forfeits 25% of the troops, so never spam the slider as a regrow trick — that lever no longer exists, and on current versions it is strictly a loss that also advertises your desperation to every neighbor watching. Work the four checks together on a single push: your bar at 45% (plateau, cheap), a western flank at 30% (soft, recovering), you the larger of the two neighbors (weight the window, re-engage soon), and a real front to conquer (no fake attack). Every check points the same way — push now into the western flank and re-engage soon — so the push is the move, and the four-check routine is the habit that keeps you from making the dump-to-zero and hoard-to-full errors on impulse. The discipline that keeps the whole thing working is the same sentence repeated: keep the bar cycled through the middle, push into soft low-fill flanks, hold the bar lower before big pushes when your cap is large, and re-engage sooner when it is small. The asymmetry is not a bonus for attackers; it is the price of a defender who leaves the plateau, and it is entirely within your control. If you keep your own bar on the better half of the curve, the “attacker is faster” problem stops being a mystery and becomes a choice you make, not a fact that happens to you, and the regrow you used to watch the opponent enjoy becomes the regrow you are producing yourself, on your schedule, with no hidden inputs and no version-specific tricks left to find.