Tactics are what you do in a moment; strategy is what you decide before the moment arrives. This guide is about the second kind - the standing decisions that make robot fighting games go better regardless of which specific machine is in front of you. Most of them transfer between games, because the underlying combat structure is shared across the genre.
Strategy one: bait rather than react
Reacting to an attack is a defensive skill. Causing the attack to happen when you want it to is a strategic one, and it is considerably more reliable.
Most machine opponents attack when you enter their range. That means their range is a switch you control: step in, and they commit; step out, and the attack misses; step in again, and you are attacking into a recovery you caused rather than one you happened to spot.
This is the difference between fighting on their schedule and fighting on yours. It is also far less demanding than pure reaction, because you know roughly when the attack is coming - you asked for it.
The technique has a name in fighting games generally: whiff punishing. The practical version is a rhythm of step in, step out, punish, rather than standing at a fixed distance and responding to whatever happens.
Strategy two: spacing is a defence
Blocking costs resources. Boosting costs resources and has a cooldown. Being out of range costs nothing at all, works against every attack including the unblockable ones, and cannot be overloaded.
Players consistently under-use distance, partly because backing away feels passive and partly because an arena is finite so retreating seems like it must run out. In practice most fights take place well inside the available space, and a step backwards at the right moment is the single cheapest defensive option available.
The exception is a timed encounter, where distance costs you the clock. Even there it is usually still correct against a telegraphed heavy attack, because the punish window afterwards recovers more than the time spent retreating.
Ad space reserved. See README.md to connect an ad unit.
Strategy three: punish correctly, not maximally
When an opening appears, the instinct is to use the biggest attack available. That is frequently wrong, and it is one of the more expensive habits in the genre.
A heavy attack has a long recovery of its own. If the opening is shorter than the heavy attack's total duration, you will still be recovering when the opponent is not - which converts your punish into their opening. A light attack that lands and leaves you free is worth more than a heavy attack that lands and leaves you exposed.
The rule: match the punish to the opening. Long openings - a completed charge, a planted slam, a deflect-induced stagger - justify a heavy. Short openings - a light attack that missed - justify a light. Guessing wrong in the conservative direction costs a little damage; guessing wrong in the aggressive direction costs a lot of health.
Strategy four: fight the crowd, not the individual
Against multiple opponents, which enemy you are attacking matters far less than what shape the group is in.
A crowd that has closed into a ring around you is unwinnable regardless of how well you aim, because a front-facing defence protects one direction and you are being attacked from all of them. A crowd strung out into a line is essentially a series of one-on-one fights, which is a situation you know how to handle.
So the strategic priority in a crowd is shape, not targets. Move in wide arcs so that faster enemies pull ahead of slower ones. Keep the arena's open space behind you rather than in front of you. Treat any movement that lets the group re-form as a mistake even if it gained you a hit.
This also reframes what a 'good position' is. It is not a position with fewer enemies near it; it is a position from which the enemies near you arrive one at a time.
Strategy five: know which fights to refuse
Not every enemy needs to die, and in continuous-spawn games almost none of them do individually.
Spending four seconds killing a heavily armoured machine while three more arrive behind it is usually a net loss even though it looks like progress. The same four seconds spent repositioning can leave you in a place where the next three fights are all favourable.
The heuristic that works: if killing something will not change your situation within the next few seconds, it is not urgent. Kill what is about to hurt you or what is blocking where you need to go. Ignore the rest until it becomes one of those two things.
Strategy six: protect the multiplier
In games where scoring is multiplicative, the multiplier is usually worth more than any individual exchange, and it is lost by taking a hit rather than by failing to land one.
This inverts the usual risk calculation late in a run. Early on, with a low multiplier, an aggressive exchange that trades a hit for two is fine. With a high multiplier, the same trade can cost more score than it gains, because everything afterwards is now worth less.
The practical consequence is that good play in these games gets more cautious as a run goes on, which is the opposite of how most people play. Most players become looser as they get further, precisely when they have most to lose.
Strategy seven: learn one enemy properly
Rather than trying to learn every enemy type at once, pick the one that kills you most often and learn it completely - its telegraph length, its attack range, what its recovery looks like, whether it can be baited.
This is more efficient than broad familiarity for two reasons. The obvious one is that it removes your most common cause of death. The less obvious one is that enemy types in these games are built from shared components, so understanding one deeply teaches you how to read the others quickly.
A good drill: in a game with a wave composition display, note which enemy is on screen when runs end. It is usually the same one far more often than chance would suggest.
Putting it together
These strategies are roughly in priority order for most players. Baiting and spacing change the most exchanges; correct punishing prevents the most sudden reversals; crowd shape decides whether the late game is survivable at all.
None of them require faster reactions. All of them are decisions, which means they can be adopted deliberately rather than developed through repetition - which is why strategy is where the fastest improvement usually comes from.
Baiting, step by step
Baiting is the highest-value strategy in the guide above and the hardest to describe abstractly, so here it is as a concrete sequence you can practise.
Stand just outside the enemy's attack range. You can find this edge by walking slowly toward it until it begins a wind-up, then stepping back - the distance at which it commits is its range, and it does not vary.
Step in. The enemy begins its wind-up almost immediately, because most arcade enemies attack on a range trigger rather than on a timer.
Step back out, during the wind-up. You are now outside the arc the attack will cover, and the enemy is committed - it cannot cancel.
The attack goes active, connects with nothing, and the enemy enters recovery. Step in and attack. The whole cycle takes under a second and you have taken no risk at any point in it.
The reason this beats reacting is that you knew when the attack was coming, because you asked for it. Reaction is a skill with a ceiling; baiting is a decision, and decisions can be made perfectly.
Reading a crowd's shape
The guide above says to string a crowd into a line. Here is how to tell whether you have succeeded, in the moment rather than afterwards.
The useful signal is how many enemies are inside your deflector's cone at once. One is a fight you are winning. Two is a fight you are managing. Three or more means the group has closed and you should be moving rather than attacking, whatever the opportunity in front of you looks like.
The second signal is where the open space is. If the largest empty region of the arena is behind you, you are kiting correctly. If it is in front of you, past the crowd, you have let them get between you and your escape - which is the situation that precedes almost every surrounded death.
Both of these are readable at a glance once you know to look, and both are more useful than counting enemies, because what matters is not how many there are but where they are relative to you.
When aggression is correct
This guide has argued consistently for patience, so it is worth being clear about the cases where the opposite is right - otherwise the advice becomes a rule rather than a judgement.
Against a shielded machine with a power meter, pressure is the answer. Blocked hits drain it and a broken plate is a large opening, so the patient approach simply does not progress.
When the multiplier is capped, additional consistency has no marginal value. Take the risks you were avoiding, because caution is no longer buying you anything.
When a timer is running, caution costs. In any game with a clock, the safe line has a price and it is sometimes higher than the risky one.
And when something is about to reach an objective you are defending, trading a hit to stop it is usually correct. Your integrity is a resource; the objective may not be.
The common thread is that patience is correct when your only limit is your own survival. As soon as something else is on a clock, the calculation changes.
Key takeaways
- Bait attacks by stepping in and out of range rather than waiting to react to them.
- Distance is the cheapest defence available: it costs nothing, works on unblockable attacks and cannot be overloaded.
- Match the punish to the opening. A light attack that leaves you free beats a heavy one that leaves you exposed.
- Against a crowd, the shape of the group matters more than which enemy you attack.
- Refuse fights that will not change your situation in the next few seconds.
- In multiplicative scoring, protect the multiplier - good play gets more cautious as a run goes on, not less.
- Bait by finding the range edge, stepping in to trigger the wind-up, stepping out, and punishing the recovery.
- Judge crowd shape by how many enemies are in your deflector cone and where the open space is.
- Patience is correct while your only limit is survival; a clock or an objective changes the calculation.
- Against shielded machines with a power meter, pressure beats patience.
Frequently asked questions
What is whiff punishing?
Deliberately causing an opponent to attack into empty space by stepping into and back out of their range, then attacking during the recovery. It is more reliable than reacting because you know roughly when the attack is coming.
Should I always use my strongest attack on an opening?
No. A heavy attack has a long recovery, so using one on a short opening means you are still recovering when the opponent is not. Match the punish to the opening.
How do I handle several enemies at once?
Change the shape of the group rather than the number. Move in wide arcs so faster enemies pull ahead, turning a ring into a line you can fight one at a time.
Is it ever right to ignore an enemy?
Frequently. If killing something will not change your situation in the next few seconds, it is not urgent. Kill what threatens you or blocks your route.
Why do my best runs end suddenly?
Often because play gets looser as a run goes on. With a high multiplier, an exchange that trades one hit for two can cost more score than it gains.
Which enemy should I learn first?
Whichever one is on screen most often when your runs end. It is usually the same type far more often than chance would explain.
How do I find an enemy's attack range?
Walk slowly toward it until it begins a wind-up, then step back. The distance at which it commits is its range, and it does not vary between encounters.
When should I stop playing patiently?
When something other than your own survival is on a clock - a timer, an objective taking damage, or a capped multiplier that consistency can no longer improve.