What a window is

A timing window is the interval during which an input counts as successful. It can be expressed in time - press within 120 milliseconds - or in space, as a region the moving object must be inside when you press. The two are equivalent, since the object's speed converts one into the other.

We express windows spatially because it makes them drawable. A player can see a zone on screen and judge their timing against it directly, which is far more learnable than an invisible interval they can only infer from outcomes.

Graded windows beat binary ones

A binary window either succeeds or fails. A graded window has bands, each scoring differently, which changes what the player is optimising.

Window typePlayer goalSkill ceiling
BinaryDo not missLow - once you stop missing, there is nothing left
Graded, 2 bandsHit the good bandModerate
Graded, 3 bandsHit the centre consistentlyHigh - accuracy keeps improving

One-Tap Candy Collector uses three bands scoring one, two and three times base value, with a precision bonus on the centre. That means a player who never misses still has substantial room to improve, which is exactly what a skill ceiling should look like.

It also means the game is legible about your performance. Hitting the outer band tells you that you were early or late by a specific visible amount, which is actionable in a way that a bare success is not.

Advertising keeps Candy Arcade free. Ads never cover gameplay or controls.

Where the measurement is taken

This detail matters more than it sounds. If the test is against the object's centre, the player should watch the centre; if it is against the leading edge, they should watch the edge. Half the object's width is a significant fraction of a narrow band.

Our tests are against the leading edge, and the leading edge is marked with a small tick so the player can see what is being measured. In testing, players who watched the centre consistently hit an outer band while players who watched the edge hit the centre with identical reaction times.

If a timing game does not tell you what it measures, it is asking you to infer a hidden rule from outcomes. That is a legibility failure rather than a difficulty.

How windows should shrink

There are two ways to make a timing window harder, and they are not equivalent.

  • Increase the object's speed. The band stays the same physical width but passes in less time, so the temporal window shrinks.
  • Narrow the band. The temporal window shrinks at the same speed.

Using both, on different schedules, produces a more interesting curve than either alone. One-Tap Candy Collector raises speed every ten collections and narrows the bands every twenty-five, so the two interleave and the player faces a new combination frequently rather than the same change repeatedly.

Both need floors. A window shorter than human reaction time cannot be hit reliably by anyone, at which point the game has stopped measuring skill.

Forgiveness mechanics are windows too

Coyote time and jump buffering are timing windows pointed the other way: instead of measuring whether your input was correct, they extend the period during which it counts.

Coyote time is a window after leaving a ledge during which a jump still registers - a tenth of a second on this site. Jump buffering is a window before landing during which a jump press is remembered, about an eighth of a second. Input buffering at maze junctions is the same idea, around a third of a second.

All three exist for the same reason: the gap between a player deciding and pressing is real, and a game that does not account for it produces failures the player cannot perceive the cause of.

Reading a window as a player

  • Find out what is being measured - leading edge or centre - and watch that.
  • Check whether the window is graded. If it is, aim for the best band rather than merely succeeding.
  • Use audio cues where they exist; they arrive slightly faster and use a spare channel.
  • Recalibrate after each speed change rather than assuming your lead time still holds.
  • Remember that forgiveness windows exist: pressing slightly late on a ledge and slightly early on a landing both work.

Common mistakes

  • Aiming to succeed rather than to hit centreIn a graded window, not missing is the floor. The scoring is in the bands.
  • Watching the wrong part of the objectHalf an object's width is a large fraction of a narrow band. Find out what is measured.
  • Not recalibrating after a speed changeThe same lead time produces a different result at a different speed.
  • Shrinking a window without a floorA window shorter than human reaction time measures nothing.
  • Hiding what is measuredIt asks the player to infer a hidden rule from outcomes, which is a legibility failure.

Practical tips

  • Watch the leading edge where that is what is measured - ours marks it with a tick.
  • In graded windows, aim for the centre band rather than for a hit.
  • Use audio entry cues as a preparation signal.
  • Recalibrate your lead after every speed step.
  • Trust forgiveness windows: late on a ledge and early on a landing both register.

Key takeaways

  • A timing window is the interval during which an input succeeds, expressible in time or space.
  • Graded windows have a far higher skill ceiling than binary ones.
  • Where the measurement is taken changes where the player should look.
  • Windows can shrink by speed or by width; interleaving both makes a better curve.
  • Coyote time and buffering are windows pointed the other way, correcting for decision latency.

Frequently Asked Questions

What is a timing window?

The interval during which an input counts as successful. It can be described as a duration or as a visible zone the moving object must be inside.

Why do some timing games have bands?

Graded bands give a player who never misses something left to improve, which raises the skill ceiling substantially over a binary hit-or-miss test.

Where should I look in a timing game?

At whatever the game measures. Ours measure the leading edge and mark it, because watching the centre makes you systematically late.

Can a timing window be too small?

Yes. A window shorter than human reaction time cannot be hit reliably by anyone, at which point it stops measuring skill.

Are coyote time and buffering timing windows?

Yes, pointed the other way. They extend the period during which an input counts, correcting for the gap between deciding and pressing.