GUIDES
OpenFront First Player Push: Size and Time Your First Commitment
OpenFront first push: the growth-curve peak, the reserve rule, the attack-cost math, and the timing window that decide whether your first offensive commitment wins or over-extends.
You have cleared a few bots, the spawn window is long gone, and a player’s border is now visible. The question is not whether to push — everyone pushes — but how much to send and when to spend it. This guide answers that one decision: sizing and timing your first offensive commitment.
The direct answer
Send your first player push at roughly 1:00-1:10 into the game, from a bar sitting near its growth-curve peak of about 42% of capacity, sized so that the troops you send exceed the target’s total army while you keep at least a quarter of your capacity back at home. Clear a weak bot first if it funds that reserve. Do not full-send on a bot, and do not make the first push against a player you cannot clearly beat.
The four numbers in that answer come from three places. The 42% peak is where the troop-growth formula tops out; the 1:00-1:10 window is when the community’s 7,000-game replay analysis shows winners actually commit; the “exceed the target’s army” rule and the “quarter back at home” reserve come from the attack-loss formula in the engine; and the “clear a bot first” line is the most common way to build that reserve safely. Everything below unpacks those four numbers so you can adjust them to your map and mode.
The reason this decision deserves its own page is that the three adjacent guides each answer a different question. The attack-ratio guide tells you what share of your army to send once you have already decided to commit to a target. The troop-bar attack timing guide tells you when to push at a given bar percentage. The spawn-kill defense guide is about surviving a first push, not making one. None of them answers the opening question: what should my first stack be, and when is the right moment to spend it? That is the gap this guide fills, and it is the decision that most often decides the next ten minutes of the game.
Before you read further, here is how to use this page. The first two sections explain the two forces that set the answer — the troop-growth curve, which tells you when a bar is worth committing, and the attack-loss formula, which tells you how much a commit costs. The middle sections turn those into a reserve rule, a sizing table, and a five-second decision checklist you can run in any game. The last two sections cover what to do when the push fails and how the numbers bend for different modes and map sizes. The single most important habit this page is trying to install is that you do not decide the first push by how full your bar looks; you decide it by whether three specific tests are green, and if they are not, you spend the next ten seconds making them green instead of clicking. That reframe — from “my bar is full so I should push” to “these three tests are green so the push is affordable” — is what separates the players the replay data shows winning from the ones who push early and small.
Why the growth curve decides your first push
OpenFront’s troop growth is not linear, and that is the whole reason the first push is a curve problem rather than a “wait until you have enough troops” problem. At the v0.35.2 tag the rate function is (10 + troops^0.73 / 4), then multiplied by a city boost of min(1 + cities * 0.06, 1.25), a dock boost of min(1 + docks * 0.08, 1.20), and finally a decay factor of (1 - troops / maxTroops). The last factor is the one that matters here. Because growth is multiplied by 1 - troops/maxTroops, the absolute rate rises while you are a minority of your capacity, peaks, and then collapses as you approach full. Worked out over a realistic early population, the peak lands at about 42% of your capacity. That is not a coincidence or a community preference — it is the mathematical top of the curve, and it is exactly the “stay near 42%” figure the community’s growth-curve guides settle on.
The practical consequence is that a bar sitting at 90% of capacity is slower to refill than one at 40%. A full bar is not a strong bar; it is a bar that has already passed the point where it grows fastest and is now bleeding growth. This reframes the first push. You do not want to wait until the bar is “full” and then dump it, because you will have wasted the fastest-growing stretch of the curve and you will be growing back from the slow end. You also do not want to push from a bar at 20%, because the stack you can afford to spare is still small relative to what a neighboring player holds. The sweet spot — and the one the 7,000-game data confirms — is to let the bar climb through the peak and commit while it is still in the fast-growing band, roughly between 40% and 55%.
There is a second reason the curve matters: the gap between your bar and the target’s bar is what the attack-cost formula actually reads. The defender’s loss is its average troops-per-tile, and the attacker’s loss is scaled by how outnumbered the push is, clamped between 0.6x and 2x the ratio of defender army to attack stack. That means the decision is really about the relationship between two numbers — your sendable stack and the target’s standing army — and the growth curve is the tool that tells you how fast you can widen that relationship. If you understand the curve, you know that pushing at the peak is not “all I have” but “the fastest rate at which my next push will be bigger than my last.” The population growth guide walks the building boosts in detail; the point for the first push is that the buildings shift the whole curve up, so a city or dock already placed moves your 42% peak in absolute troops, letting you send a bigger first stack at the same bar percentage.
The reserve rule: a push you can actually afford
A push is only as good as the bar you keep behind it. The single most common first-push failure in the community threads is the player who sends everything, wins the exchange, and is left at a trickle of troops while the target refills — turning a small win into a slow loss. The fix is a reserve, and the reserve is not an optional safety margin, it is the thing that makes the next twenty seconds of the game worth the push in the first place.
Concretely, the reserve rule is: after your first push, your home bar should still be at least a quarter of capacity. Why a quarter? Because the growth curve means a bar at 25% refills far faster than one at 10%, and because the attack-cost formula means that when the target counters, you want a second stack that is still larger than the counter. If you sent 60% of your capacity and the target answers with a 20% counter, you can answer back from the 40% you kept. If you sent 95%, you answer back from 5%, and the counter outruns your refill. The “100% of my maximum or more while I still have another 30% at home” line from the community’s growth-curve guide is this exact idea: the push is bigger than the bar, which is only possible because the “bar” you are measuring is the sendable portion and the home portion is the reserve that keeps the curve on its fast side.
The reserve also has a strategic value that pure math does not capture. A visible reserve is a deterrent. A player who sees that you cleared a bot but kept a large standing army is less likely to full-send into you, because they cannot be sure they can beat the counter. That is not a formula, it is a read on the opponent’s decision-making, and it is one of the reasons the winners in the replay data push slightly later (around 1:06) than the losers (around 1:00) while still committing bigger. The later push is often the one that had time to build the reserve. So the reserve rule does double duty: it protects you mechanically through the attack-cost formula, and it shapes what the opponent believes is safe to attempt.
One refinement: the reserve is per-neighbor. If you have two player borders, the quarter you keep has to be split between the two, which is why two-front openings want a bigger reserve than a single-border opening. The multi-front defense guide covers holding two borders once the game is underway; for the first push the rule is simpler — if you are facing two live player borders, do not make the first push until the reserve that covers both borders is at least a quarter of capacity, or pick the weaker of the two and commit to only one of them.
Sizing the first stack with the attack-cost math
Once you accept the reserve rule, the remaining question is how big the sendable portion actually needs to be. The v0.35.2 attack function gives the attacker’s troop loss as mag × clamp(defender.troops / attackTroops, 0.6, 2) × (0.463 + 0.0039 × density), where density is the defender’s average troops-per-tile, mag is the terrain magnitude (plains 80, highland 100, mountain 120 in the v0.35.2 default table), and a human attacker against a bot defender gets mag multiplied by 0.7. The defender’s loss is simply its average troops-per-tile per attack. Two things fall out of that formula that change how you size the first push.
The first is the clamp. The outnumbered ratio defender.troops / attackTroops is clamped to 0.6 on the low end and 2 on the high end. On the high end, being badly outnumbered does not keep getting cheaper for you — once you are more than about 1.67x the defender’s army, the clamp has already floored the ratio at its benefit, and sending more troops costs you more in absolute numbers than it saves. On the low end, if your stack is smaller than the defender’s army, the ratio climbs toward 2 and your per-tile loss rises. So the “cheapest” attack is not the biggest attack; it is an attack in the band where you outnumber the defender by a comfortable margin but have not yet passed the point where the extra troops are pure waste. In practice that band is roughly 1.5x to 2x the target’s standing army.
The second thing is the density term. The defender’s cost scales with how packed its land is. A target sitting on 80 tiles with 9,000 troops (112 per tile) is more expensive to hit per tile than the same 9,000 troops spread over 150 tiles (60 per tile). This is why “encircle the bot and hit it from several sides” is not just a map-tip — it is a way to reduce the effective density you pay, because you are spending the same stack against a thinner front. The table below shows the attacker loss for a plains engagement at v0.35.2, so you can read your own numbers against it:
| Your send | Target army | Outnumbered ratio | Attacker loss (bot, plains) | Attacker loss (player, plains) |
|---|---|---|---|---|
| 5,000 | 9,000 | 1.8x | ~91 | ~130 |
| 7,000 | 9,000 | 1.3x | ~65 | ~92 |
| 9,000 | 9,000 | 1.0x | ~50 | ~72 |
| 11,000 | 9,000 | 0.8x | ~41 | ~59 |
| 15,000 | 9,000 | 0.6x (clamped) | ~30 | ~45 |
Read the table against the reserve rule: the first column where your remaining home bar is still a quarter of capacity and the attacker loss is already low is your send size. For a 9,000-troop bot, that is usually the 11,000-15,000 send, not the 5,000 send. The 5,000 send wins the exchange on paper but spends a bar you then have to regrow slowly, and it risks the low-end clamp if the bot is actually larger than your scout suggested. The land combat guide covers the per-tile mechanics in depth; for the first push the table above is the sizing tool.
Timing the commitment: when the window closes
The 7,000-game replay analysis found that winners make their first player push at about 1:06 and losers at about 1:00, with the winners committing a roughly 34% larger stack (about 17,300 versus about 12,900 in the sample). The six-second gap is small, but it is the difference between a push made with a reserve and a push made the instant the spawn window expires. The reason is that the first ten to twenty seconds after the window are spent doing the unglamorous work that makes a bigger push possible: clearing the first weak bot, taking the first neutral tile you want, and letting the bar climb through the growth-curve peak. A player who pushes at 1:00 is usually pushing before that work is done, which forces a smaller stack from a lower bar.
The timing also interacts with the spawn phase itself. At the v0.35.2 tag the random-spawn phase is 200 ticks, which at 10 ticks per second is about 20 seconds; singleplayer is 100 ticks, about 10 seconds; and the team and 2v2 modes use longer windows. The default public-FFA immunity is 50 ticks, about 5 seconds, and during it a human cannot full-send at another player, though it can still expand into neutral land and take bots. So the “window” that matters for your first push is not the immunity — that is gone within the first few seconds — but the practical window where the neighbor has not yet built a reserve of its own. Push too late and the neighbor has had time to sit at its own 42% peak and commit a bigger stack than you. Push too early and you are the one without a reserve.
The practical timing test is therefore: push when your bar is in the fast-growing band (40-55%), your reserve is at least a quarter of capacity, and the target’s visible bar is at or below yours. If all three are true near 1:00-1:10, commit. If the target is clearly larger, do not make the first push on it — clear a bot or take neutral land to widen the gap, and re-test in ten seconds. The tall versus wide guide covers when a safe economy should start taking more land; for the first push the timing rule is the same shape — take the land while you are growing, not after you have stopped. The worst first push is the one made at 90% of a bar that has already passed the peak, because you are committing your slowest-growth troops into a target that is still growing.
A decision framework for the first push
The three numbers (curve peak, reserve, target army) collapse into three questions you can run in about five seconds, before you click. Ask them in this order, and if any answer is red you do not push yet.
The first question is the affordability test: can I clear the target at my current stack and still have a quarter of capacity back at home? Work it against the table above. If the target is a 9,000-troop bot and you can only send 6,000, you are in the low-end clamp and the answer is red — the exchange is not cheap enough. If you can send 12,000 and keep a quarter home, the answer is green. This question uses the attack-cost formula, and it is the one that stops the most common first-push failure, which is over-sending on a bot.
The second question is the opportunity test: what is the next twenty seconds worth if I do not commit? If the answer is “a weak bot I can clear at a 2x ratio” or “a neutral chokepoint that opens a second growth ring,” then not pushing is worth more than pushing, and you should take the land first. The first push is a commitment of troops that will be gone for the rest of the exchange, so the alternative use of those troops has to be genuinely worse before the push is the right call. This is the question the growth curve answers: if you are still in the fast-growing band, the land you could take in the next twenty seconds is usually worth more than a marginal push now.
The third question is the counter test: after the push, will my home bar be able to beat the counter? Look at what the target will have left. If the target keeps a reserve of its own, your home bar has to be able to beat it in the follow-up. If you sent 60% and the target sent 40% and kept 30%, your remaining 40% has to beat their 30% counter — and it does, but barely, and only if your terrain is as good as theirs. If the terrain is worse or the target has a dock, the counter outruns you and the first push has just traded your growth for a border you cannot hold.
These three questions are the tool. They turn the first push from a gut call into a checklist you can run every game, and they are the reason the winners in the replay data commit slightly later and bigger — they are the players who passed all three questions before clicking. Write the three questions at the top of your notes if you have to; the first push is the one decision of the game where a five-second check is cheap insurance.
Scenario A — clearing a weak bot to fund a player push
Setup: random-spawn World map, about 20 seconds in, you sit at 90 tiles. A bot of 70 tiles holds 8,000 troops (about 114 per tile) on a plains border. A player of 80 tiles holds 11,000 troops (about 138 per tile) on your other plains border. Your bar is at 44% of capacity with no buildings yet. The question is whether to hit the bot or the player first.
Run the three questions on the bot. Affordability: your capacity at 90 tiles is roughly 120,000; a quarter home is about 30,000, so you can send up to about 90,000. The bot is 8,000, so the outnumbered ratio is enormous and the attacker loss per attack is minimal (the clamped ratio sits at the 0.6 floor). Clearing 8,000 troops against 90,000 you can send is trivially cheap — the bot will fall in a handful of attacks with near-zero loss. Green. Opportunity: the alternative to clearing the bot is sitting on the player at 11,000 with an 8,000-troop bot still between you and a second growth ring, so the bot is the better land. Green. Counter: a bot does not counter, so the reserve is preserved for the player. Green. All three are green, so you clear the bot first.
Now run the same three on the player, after the bot is gone and you have taken its 70 tiles. Your capacity is now higher (more tiles), the bar has climbed back into the fast band, and the player is still at 11,000. Affordability: you can now send a stack of 20,000+ and keep a quarter home; 20,000 against 11,000 is a 1.8x ratio, in the cheap band, with attacker loss around 45-60 per attack on plains. Green. Opportunity: the player is now your only live border and the land they hold is worth more than the neutral tiles left. Green. Counter: the player keeps a reserve, but your 40% home bar beats their expected 30% counter on equivalent terrain. Green. You make the player push now, near 1:00-1:10, from the bigger bar the bot-clearing bought you. The numbers show why the sequence matters: hitting the player first from the 90-tile position means sending 15,000 against 11,000 with no second ring, a thinner exchange; clearing the bot first buys you a bigger bar and a 1.8x ratio, which is the 34% gap the replay data attributes to winners.
Scenario B — taking on an equal-strength player near the one-minute mark
Setup: World map FFA, about 55 seconds in, you sit at 95 tiles and the only player border is an equal-strength player at 95 tiles with 12,000 troops (about 126 per tile) on a highland border. Your bar is at 47% of capacity and you have no city yet. The temptation is to full-send, because the bar looks strong and the player looks even. Run the three questions.
Affordability: a quarter of your capacity at 95 tiles is about 31,000, so you can send up to about 92,000. Against 12,000 that is a 7x+ ratio — far past the clamp, where the extra troops are pure waste and the attacker loss per attack is only marginally better than at 2x. The cheap send here is 18,000-24,000 (a 1.5x-2x ratio), not 90,000. So affordability is green, but only if you send the cheap 2x stack and not the full bar. This is the over-send failure: the bar says “send everything” and the formula says “send 2x.” The reserve rule forces the 2x send.
Opportunity: the highland border costs more per tile than plains (mag 100 versus 80), so the exchange is more expensive than the table suggests for plains. The alternative use of the troops is to take a neutral highland ring and build a city to shift the growth curve up. If the player is not actively pushing at you, the opportunity cost of the push is real, and the answer leans red — hold and build. Counter: the player keeps a reserve on the same highland; your home bar has to beat their counter over highland, which is a worse trade than the plains table. Red-leaning.
The verdict: do not make the first push on this player at 55 seconds. The correct move is to take a neutral ring, build the city (which moves your 42% peak up in absolute troops), and re-run the three questions in about fifteen seconds. By then you will be at a bigger bar, on a city-boosted curve, and the player will either have pushed at you (in which case you defend, see the spawn-kill defense guide) or will still be holding (in which case your 2x stack is now bigger and the exchange is cheaper). The scenario is a study in why “equal strength” is not the same as “even exchange”: terrain and reserve turn an equal bar into a losing push if you full-send, and a winning one if you send the cheap 2x and keep the reserve.
When the first push fails, and how to counter it
The two ways the first push goes wrong are over-sending on a bot and full-sending on an equal player, and both leave you in the same bad state: a won exchange you cannot follow up on, because your home bar is too small to beat the next counter. Recognizing that state the moment you are in it is what saves the game, because the fix is a defensive posture, not a second push.
If you over-sent on a bot and are left at 10% of capacity while the bot’s neighbor (a player) refills, the counter is to stop pushing and consolidate. The attack-cost formula favors the defender, so sit on the border you just took, let the bar regrow through the fast band, and do not initiate again until the reserve rule is satisfied. The temptation is to “finish the job” with a second push while the bar is low; that second push is the one that loses the game, because it is made from the slow end of the curve and the target has had time to regrow. The community’s “don’t just keep attacking, let your troops build up in between” advice is this exact correction.
If you full-sent on an equal player and the exchange is even or worse, the counter is the same shape but faster: fall back to the border you can hold, and if the player has committed their reserve to the push, their home bar is now low — which is the one moment where a small counter-push is viable, because you are now the one with the cheap ratio. This is the “they won’t forget” failure from the community threads: a player who full-sends at you early is exposing their home bar, and the player who holds a reserve is the one who gets the counter. So the defensive counter is also an offensive one, just delayed until the target has made itself cheap.
The general rule is: the first push is the only push of the opening you are allowed to lose cheaply, because the information it buys (the target’s real army, its reserve, its terrain) is worth more than the troops. A first push that over-extends is not a lost push, it is a spent bar you have to rebuild, and the game is still winnable if you stop pushing and let the curve work for you. The threat assessment guide covers reading whether the next counter is coming at all; for the immediate aftermath the rule is simple — no second push until the reserve rule is green again.
Mode and map adjustments
The three questions and the table hold across modes, but the version-sensitive numbers change them. In 2v2 and Team modes the spawn window is longer than the 20-second random-spawn phase, which buys you more time to build the reserve before the first push; that means the timing test (your bar in the 40-55% band with a quarter home) is the same, but the clock to reach it is slower, so the first push tends to come later and bigger. In singleplayer the 10-second phase and the bot-only early game mean the “first player push” is really the “first hard-bot push,” and the affordability test is almost always green, so the binding constraint is the opportunity test — take the land while you are growing rather than polishing a bot exchange.
Map size changes the density term, which is the most map-sensitive number in the whole framework. On a compact map the borders are short and the defender’s troops are spread over fewer tiles, so the density per tile is higher and the exchange is more expensive per tile; the first push wants a bigger 2x margin to stay in the cheap band. On a large map the borders are long and the target’s army is spread thin, so the density is lower and the same 2x stack costs less; the first push can be made earlier and smaller. The map strategy guide and the mode-specific opening guides (the World-map FFA and 2v2 openings) carry the per-map details; the framework here is the part that does not change with the map.
Two final adjustments. First, buildings: a city or dock placed before the first push moves your growth-curve peak up in absolute troops, so the same 42% bar carries a bigger stack and the affordability test passes at a smaller bar percentage. This is the single highest-leverage move in the opening and the reason “buy the city early” is recurring advice. Second, terrain at the border: if the first border is highland or mountain, the mag is 100 or 120 instead of 80, and every number in the table scales up proportionally; the cheap 2x band still applies, but the absolute stack you need is bigger. The table is the plains baseline; multiply the attacker loss by 1.25 for highland and by 1.5 for mountain to get the adjusted send size. None of these adjustments changes the three questions; they only change the numbers you plug into them.
Related content
Use OpenFront v34.3 costs, troop growth, route payback, and stop signals to choose City, Port, Factory, defense, or a liquid reserve without copying a fixed script.
- OpenFront AFK Teammate Takeover: Absorb a Disconnected Ally for Zero Troop Loss
A Team-mode decision guide for what a teammate's disconnect does to your win/loss position, when absorbing their land is free, when it is a trap, and what the 30-second mark and the ranked 2v2 rules actually change.
Sep 23, 2026
- How Nations Aim Their Nukes: Smarter Targeting, No Wasted Blasts
A strategy guide for the upcoming AI Nation nuke targeting change: Nations now compute how much of a blast hits enemy land before firing (atom 25%, hydrogen 40% minimums), they skip a spot where another player's nuke is already landing, and they never nuke a target that would hit friendly land. What changes in your decision, the coverage gate, the anti-double-nuke rule, the failure mode and the mode/map adjustments.
Oct 8, 2026
- Alliance-Break Timing: When 30 Seconds Is Worth the Trade
Decide whether to break a live alliance now, wait for it to expire, or renew it, using the 30-second traitor window, the 0.5 defense and 0.8 speed debuffs, and the -100/-40 relation drops in OpenFront v0.34.
Sep 30, 2026