Should You Recall or Let an Attack Finish?

A recall in OpenFront is a deliberate exit from an attack you are already running, and it always carries a price if the target is a human player. The single number that drives the decision is the flat twenty-five percent troop loss the game applies to any attack you cancel against a player. You either finish the fight and accept whatever a contested battle costs, or you pull the troops home and pay a guaranteed quarter of the force you committed. The right move depends on whether the fight you are in is likely to cost you more than that twenty-five percent. This guide gives you the framework for choosing between those two exits, the exact rules the engine enforces, and the land-versus-sea difference that lets a sea attack be cancelled for free once it has left your own waters.

The 25% recall rule and when it binds

When you order a recall the game does not refund the troops instantly. It sets a retreat, and the engine keeps a fixed penalty of twenty-five percent of the total troops in the attack. That penalty is the value stored as attackPercent in the retreat execution, and it is a flat percentage of the force you sent, not a function of how the fight is going. The penalty only binds when the attack’s target is a player rather than a bot. The code branches on whether the target is a bot: if the target is a bot, the retreat malus is set to zero and your full force comes home; if the target is a human player, the malus is set to twenty-five percent and a quarter of the troops are lost on the way back.

This distinction is the first and most important thing to internalize. Against an AI nation a recall is free in the sense that no troops are destroyed; the engine simply sends the whole force home with a zero malus. Against a human opponent the recall is a real cost: you lose exactly twenty-five percent of the troops you committed, no matter how badly the fight is going. The cost is the same whether the attack was about to win or about to be annihilated. A recall is therefore a capped-loss tool: it converts an uncertain, potentially large battle loss into a certain, fixed twenty-five percent loss.

The rule applies to every attack category the engine runs through the same retreat path: land forces, boats, and transport ships all share the same twenty-five percent malus when the destination is a player. The land and sea difference is not in the penalty percentage but in when the penalty is charged, and that is covered in its own section below. For now the working model is: a recall against a player costs a flat twenty-five percent of the committed troops, and a recall against a bot costs nothing. Notice that the percentage applies to the troops inside that one attack order, the force the engine tracks for the attack, not to your entire standing army. A small fifty-troop scout and a four-hundred-troop battalion each lose a quarter of themselves, so the absolute number of troops at risk scales with how much you committed to this single attack. That is why committing a very large force to one recall-able attack is more dangerous in absolute terms even though the fixed percentage never changes, and it is the reason the later scenarios size their forces deliberately instead of dumping the whole army into one order.

Why a recall is not the same as a losing fight

The most common mistake is to treat a recall as the default “I am losing” response and to hit the exit the moment the attack turns ugly. That is usually the wrong call, because a losing fight and a recall are not the same cost profile, and a fight that looks dire is often cheaper to finish than to abandon. In a contested battle your losses depend on the attack-to-defense ratio you are actually facing. If you are attacking with a comfortable numerical advantage the fight may lose you well under twenty-five percent and still take the tile, which is a strictly better outcome than paying twenty-five percent to walk away from nothing.

The recall only makes sense when the fight you are in would cost more than twenty-five percent. That happens when you are massively outmatched, when a second enemy flank arrives after you committed, or when the target has just reinforced to a level that turns your advantage into a deficit. In those cases finishing the attack means losing the majority of your force, while a recall caps your damage at a quarter. The decision is a comparison: estimate the loss you would take if the attack ran to the end, and compare it to the fixed twenty-five percent. If the projected fight loss is above twenty-five percent, the recall is the smaller cost and the correct exit. If the projected fight loss is below twenty-five percent and the capture is still on the table, finishing is the better trade even though it is a fight you do not enjoy.

This is why the recall is a decision and not a panic button. It is the move you make when the alternative is a catastrophic loss, and it is the move you should not make when the alternative is a winnable fight. One subtlety sharpens the comparison: because the twenty-five percent is charged on the troops you committed to that attack, the size of your commit decides whether a “bad” fight is tolerable. A fight that would lose half of a four-hundred-troop force is a two-hundred loss you should abandon, but the same half against a fifty-troop probe is a twenty-five loss that may be worth taking for the information it buys. The ratio is the driver and the size of the commit sets the scale, so the two numbers have to be read together, not the ratio alone. Holding that comparison in mind, the rest of this guide is about reading the situation fast enough to make that comparison while the window to act is still open, because the exit is not instant and the next section explains exactly how long you get.

The land-versus-sea asymmetry

The same twenty-five percent malus behaves very differently on land at sea, and this difference is the single most useful tactical fact in the whole recall system. The engine charges the retreat penalty only when the attack’s current tile is still owned by the attacker. On land the origin tile of the attack is always your own tile, so a land attack is always sitting on your land when you cancel it, and the twenty-five percent malus always applies. There is no circumstance under which a land recall is free against a player: the origin is yours, the tile check passes, and a quarter of your land troops are destroyed.

At sea the rule flips into a genuine option. A boat or transport only carries the twenty-five percent malus while its current tile is one you own. The moment the vessel has moved out of your waters and is sitting on an enemy or neutral tile, the tile check fails and the malus is set to zero, so the full force comes home untouched. In practical terms this means a sea attack can be cancelled for free after it has left your own coast. You can send a boat toward a distant enemy coast, let it cross the boundary, and then recall it with no troop loss. The enemy never captures anything, you lose nothing, and you have paid only the time it took the boat to travel.

This asymmetry is the foundation of a whole class of sea decisions. It turns a sea attack into a low-cost probe: you can test whether an enemy coast is defended, watch how a garrison responds, and abort the moment the response is stronger than expected, all without committing any troops permanently. It also means the timing of your sea cancel matters in the opposite direction to a land cancel: for a boat you want to cancel after it has left your water, not before. Cancelling a transport still sitting in your own harbor is the one sea case that does cost twenty-five percent, because the vessel is still on your tile. Land always costs, sea costs only while you are home, and that is the entire shape of the asymmetry. There is one edge case worth pinning down because it is where players get the rule backwards: a transport that has already crossed and is sitting on an enemy or neutral tile is free to cancel even after it has landed a force, as long as the force is counted as the attack and is still on a non-attacker tile. The moment, however, that you cancel while the vessel is still in your own port, the vessel is on your tile and the full twenty-five percent is charged. The free exit is strictly a property of having crossed the border, which is exactly why the sea probe described later waits until the boat is off your own coast before it pulls the cancel trigger, and it is the reason a sea attack should never be recalled from the harbor in a hurry.

The 20-tick cancel window and what the enemy can do with it

A recall is not instant, and the delay is long enough to change the decision. When you order a retreat the engine does not turn the force around immediately. It applies a fixed cancel delay of twenty ticks before the retreat actually executes. At the engine’s tick rate of one hundred milliseconds per tick that is a full two seconds of continued advance before the troops start coming home. For two seconds your attack keeps moving toward the target, keeps closing the distance, and if it is close enough to the destination at the start of that window it may still complete the landing or the capture before the recall takes effect.

This has two consequences you should plan around. The first is that the recall window is also a risk window for the defender. The enemy can see that your force has stopped advancing and started to turn, and those two seconds are an opening to reposition a defense, shove a counterattack into your exposed flank, or push a second force at the route you have just vacated. The engine gives them that window for free, so a recall you issue against a prepared player is a recall they can punish. The second consequence is that the window is a hard decision deadline for you. Because the troops keep moving for two seconds after you click recall, a recall issued at the last possible second is a different, more expensive action than one issued early. The earlier you decide, the more of the two-second advance happens before the force commits to the tile it is about to reach.

Treat the twenty-tick delay as the boundary of the decision, not as a grace period. If the fight is clearly lost the moment you notice it, recall immediately so the two seconds of continued advance land you as close to the safe side as possible. If you wait until the last second, the advance may still complete the capture and the recall will only save you from the follow-on fight. The practical rule is to make the recall decision on the first clear signal that the fight will cost you more than twenty-five percent, and to stop treating the delay as something you can use to “maybe finish it instead.” The delay is a trap for the indecisive, and the next section gives you the framework that keeps you out of it. In short the twenty-tick window is the difference between a recall you control and a recall that controls you, and reading it correctly is the only way the flat twenty-five percent becomes a tool instead of a tax you pay by mistake.

When to recall: the decision framework

The framework reduces the decision to one comparison and two filters. The comparison: what would this fight cost me if I let it finish, and is that cost above or below twenty-five percent of the force I committed? The two filters: is the target a player (recall costs twenty-five percent) or a bot (recall is free), and is the attack on land (always costs) or at sea and already out of my water (free)? Work the three answers and the exit follows directly.

When the target is a bot the answer is almost always recall early, because a free recall dominates a fight against an AI that you may as well disengage from the moment the tile is not worth the travel. When the target is a player on land, the cost is fixed at twenty-five percent and the only question is whether the fight would cost more. If yes, recall; if no, finish. When the target is a player at sea, the cost is twenty-five percent only while the vessel is still in your own waters and zero after it has crossed, so the dominant strategy is to let the boat cross and then cancel for free unless the capture is clearly winnable.

The table below summarizes the decision for the common cases. Use it as a lookup, but always sanity-check it against the live situation, because the one number it cannot see is whether a second enemy force is about to arrive.

Situation	Target	Cost if you finish	Recall cost	Correct exit
Land attack, clearly outmatched	Player	Above 25%	25%	Recall now
Land attack, winning or even	Player	Below 25%	25%	Finish
Sea probe, boat past your coast	Player	N/A (abort)	0%	Cancel free
Sea probe, boat still in your harbor	Player	N/A (abort)	25%	Wait to cross, then cancel
Any attack vs AI that is not worth it	Bot	Travel + fight	0%	Recall free
Alliance forms mid-attack, target now ally	Player	Continued fight	25%	Let auto-retreat run

The framework is only as good as your estimate of the fight loss, and the estimate is where most players err. Overestimating how bad the fight is makes you recall winnable attacks and give up captures you would have won; underestimating it makes you finish attacks that should have been abandoned and lose the majority of your force. A useful discipline is to estimate the fight loss twice: once assuming the defense stays exactly as it looks now, and once assuming the defense gains a meaningful reserve. If the “static” estimate is below twenty-five percent but the “reinforced” estimate is above it, the fight is a coin flip on information you do not have, and the safe move is to probe first rather than commit a large force blind. The scenarios in the next two sections walk one of each, with the numbers laid out, so you can calibrate your own estimate against a worked example before you apply the framework live.

Scenario: an overextended land assault

Consider a concrete land case with the numbers written down. You have committed four hundred troops to a two-tile push at a defended enemy city. The city has roughly two hundred and fifty troops between garrison and a nearby reserve, and your four hundred looked like a comfortable two-to-one at the moment you launched. But the launch was on your own initiative, with no trade-off forcing you to take it, and while your attack closed the distance a second enemy force of three hundred troops came around from the flank and joined the defense. The ratio you are actually facing now is four hundred against five hundred and fifty, a clear disadvantage, and the fight is going to cost you well more than a quarter of your force if you let it run to the end.

Apply the framework. The target is a player, so the recall costs a fixed twenty-five percent of four hundred, which is one hundred troops lost with three hundred coming home. Now estimate the finish: at four hundred to five hundred and fifty the battle would destroy the larger share of your smaller force, realistically losing you three hundred or more of the four hundred and leaving you with a handful of stragglers and no tile. The finish costs three hundred or more; the recall costs exactly one hundred. The recall is the smaller loss by a wide margin, so the correct exit is to recall now, take the one hundred loss, and bring three hundred troops back to a position where you can choose a better fight.

The assumption doing the work here is the flank arrival. Without the three hundred arriving the ratio stays near two-to-one, the fight would lose you well under a quarter, and the correct call flips to finishing and taking the city. The whole decision hinges on that single fact, which is exactly why the framework keeps the “is a second force arriving” check outside the table. When you are reading a live board the thing to look for before you commit to either exit is whether the defense is static or about to grow, because a static defense you already outnumber is a fight worth finishing even at a bad-looking instant, while a defense that is about to triple is a fight you should abandon the moment you see the second force commit. There is also a tempo cost to consider: the two hundred troops you lose to the recall are gone for the rest of the match, so a recall is only cheap if the three hundred you save can do more than hold their own. If recalling leaves you with a force that can neither recapture the route nor defend your core, the “smaller loss” is still a loss that changes the whole board, and the correct exit may be to hold the line and finish the fight instead of withdrawing into weakness.

Scenario: a sea probe you can cancel free

Now the sea version, which uses the asymmetry in the previous section. You are considering whether to land on a distant enemy coast that you have not scouted, so you send a transport ship carrying three hundred troops toward it. The transport is the right tool for this because you want to be able to abort. As the transport moves you watch the enemy coast: for the first few ticks there is no visible garrison, which is encouraging, but you cannot be sure the defense is not simply hidden behind the tile. You do not want to commit three hundred troops to a landing against a coast you have not actually measured.

The framework changes the decision here in a way land never does. The target is a player, so if the transport were still sitting in your own harbor and you cancelled now the cost would be twenty-five percent, seventy-five troops lost. But the transport has already crossed out of your waters and is on a neutral or enemy tile, so the tile check fails and the malus drops to zero. You can recall the transport now and bring all three hundred troops back home, untouched, having paid nothing but the travel time. The probe is now a free information buy: you saw how the coast responds, you know roughly how strong the garrison is, and you have not lost a single troop. You can then either commit a properly-sized follow-up attack if the garrison is weak, or simply not attack at all if it is strong.

The assumption that makes this work is the crossing. The free cancel only exists once the vessel is off your tile; the moment before that the cancel costs seventy-five troops, which is why the play is to let the boat cross the boundary before you pull the trigger. And the counter to this play is that the enemy has now seen your transport arrive and withdraw, which tells them you are assessing that coast. Do not run the same probe three times at the same spot; use it once to measure, then make the commitment or move on. The sea probe is the one attack you can afford to run on purpose and walk away from for free, and that is what separates a disciplined sea operation from a land raid that has to be all-or-nothing. There is one more reason the probe is powerful: because it costs nothing in troops, you can run it early in the match when you have the most to learn and the least to lose, and bank the information for the mid-game when a properly-sized follow-up is actually worth committing. A land probe you would pay twenty-five percent for you might not run at all in the opening; the sea version is free enough to run on almost every interesting coast, which makes the early map far better scouted by the time the real fights begin.

Failure modes, counterplay, and mode or map adjustments

The recall has clear failure modes, and each one has a counter you can apply while the attack is still live. The first failure is the late recall: you wait until the last second of the twenty-tick window so that the continued advance still lands the force on the destination tile and completes the capture, converting a “free” recall into a capture you did not want. The counter is to make the recall decision on the first clear signal, not at the boundary, so the two seconds of advance happen over open space rather than over the tile you are trying to avoid. The second failure is the recall that invites a counterattack: you pull out and the enemy, seeing your force turn, pushes a second force into the route you vacated or hits your exposed flank during the two-second window. The counter is to recall toward a defended line or behind a tile you can immediately recapture, so that the two seconds the enemy has to pounce are spent against a line that is ready for them rather than an empty border.

The third failure is the false probe: you send a sea boat to measure a coast, cancel it for free, and then the enemy reinforces that exact coast because they watched your boat turn back. The counter is to treat the probe as one-shot intelligence; if the coast is worth attacking at all you attack it with the follow-up on the same turn you decide to commit, not three turns later when the garrison has tripled. The fourth failure is the ratio misread, the one that costs real troops, where you under-estimate the fight loss and finish an attack that the framework would have told you to abandon. The counter is the discipline of writing the ratio down before you commit: name your force, name the visible defense, and if the defense can grow, assume it will and compare your loss to twenty-five percent with that growth already in the number.

Mode and map shape change which exit is worth taking. In No Alliances mode you cannot rely on an alliance forming to bail you out of a bad attack, so your only exits are the twenty-five percent land recall or the free sea recall, and you should plan your attacks with that narrower toolbox from the start. On island and water-heavy maps the sea asymmetry becomes the dominant tool: most of your attacks are sea attacks you can cross and cancel for free, so probe freely and commit only when the measurement justifies the follow-up. On a Cold War board the land recall is your intel tool: a land probe you are willing to pay twenty-five percent to run tells you how defended a neighbor is before you decide whether a full commitment is worth the cost. In every case the shape of the map decides whether the free exit is on the table, and the framework in the earlier section decides whether taking it is the smaller loss than fighting.

Putting the recall decision into your attack plan

Bring the pieces together into a single habit you can run on every attack. Before you launch, name the three inputs the framework needs: is the target a player or a bot, is this land or sea, and what is the visible defense ratio? That naming costs you nothing and it puts the decision inside your head before the fight starts, so the recall is a planned branch of the plan rather than a reaction. Then define the trigger that switches you from finishing to recalling. The cleanest trigger is a fixed line: if the fight would cost me more than twenty-five percent of this force I abandon it. Because the recall cost is exactly that twenty-five percent, the trigger is simply the point where the fight loss crosses the recall cost, and a fixed line is something you can check in one second while the board is moving.

Once the trigger is set, the execution is a discipline of timing. For a land attack against a player, the moment you see the defense grow past the point where the fight exceeds twenty-five percent, recall immediately and accept the one-quarter loss; do not negotiate with the last second. For a sea attack, let the vessel cross your own waters first and then cancel free unless the measurement says the capture is winnable; the crossing is what unlocks the zero-cost exit, and skipping it turns a free probe into a seventy-five-troop mistake. For any attack against a bot, recall the moment the tile stops being worth the travel, because the free recall against an AI dominates almost every fight you would otherwise lose troops to.

This is the whole system, and it is short enough to run under pressure. A recall against a player is a flat twenty-five percent loss, a recall against a bot is free, land always pays and sea pays only while it is still home, and the twenty-tick window means the decision has to be made early or not at all. The recall is not a sign of defeat; it is a capped-loss tool that lets you trade a small, certain cost for the chance to choose a better fight later. Use it the way the rest of this site uses attack ratio and threat assessment: as a measured decision made on the numbers you can actually see, rather than a reaction to a fight that is already over. When the next section of your plan is a fight you cannot win, the recall is the exit you take, and now you know exactly what it costs before you take it. See the attack ratio guide for the fight-loss side of the comparison and the transport landing guide for the sea operations this recall rule closes the loop on.