Speed Breakpoints

A breakpoint belongs to a deadline

A Speed breakpoint is the SPD needed to fit a chosen number of normal turns into a chosen amount of battle time. Change either part of that question and the answer changes. A number such as 134 is shorthand for a timing problem, not a property that makes every build good.

This is the next step after Speed & Action Order. Here we use Action Value, or AV, to derive a few deadlines. We start at battle time zero with a full 10,000 action distance, constant positive SPD, and no advance, delay, extra turns or wave transition. Every normal turn resets that distance. These assumptions matter as much as the answer.

Turn a deadline into an inequality

One natural turn takes 10,000 ÷ s AV, where s is constant in-battle SPD. The nth turn therefore arrives at 10,000 × n ÷ s. To reach it strictly before a window of W AV ends:

10,000 × n ÷ s < W, so s > 10,000 × n ÷ W.

The right-hand side is the exact mathematical boundary. Equality puts the action precisely on the deadline; it does not put it before the deadline. We use a strict inequality so the examples do not depend on a cycle-marker tie. This is a conservative timing test, not a claim that every tied action is lost in the game.

134 and 160 answer different questions

The familiar Memory of Chaos timing model gives its opening cycle 150 AV and the next cycle 100 more. That makes the first two cycles a 250 AV cumulative window. It does not give every cycle 150 AV. These examples describe one uninterrupted timing sequence; a new wave or a mode-specific reset needs its own check.

134 is a common rounded player target. At 133 SPD, the second turn arrives at about 150.38 AV, just beyond the opening window. At 134 it arrives before it. The exact boundary is still 133⅓; rounding that boundary down to 133.3 would miss it.

160 is the exact four-in-250 boundary. Its turns arrive at 62.5, 125, 187.5 and 250 AV. Saying “160” does not prove a strict-before result. Effective 160.1 gives a small margin; effective 161 gives a larger one. Neither margin is a new universal breakpoint. A rounded stat display alone is not enough to resolve a close tie.

Four turns in 250 AV also does not mean two turns in every later cycle. Count the actual action times. The 201 example illustrates the same calculation at a different goal; it is not a recommendation to spend enough Relic stats to reach it.

Use the time the encounter actually gives you

Suppose a hypothetical objective gives you 200 AV from a fresh start and you want three turns. The boundary becomes 30,000 ÷ 200 = 150 SPD. At 151, turn three lands at about 198.68 AV. That is an original timing example, not a rule assigned to a particular mode.

For another mode, establish its action window and reset rules first. For a buff window, check whether its duration is measured in elapsed AV, the holder’s turns or another unit’s turns. A two-turn buff is not automatically a 150 AV window. For a short fight, the important deadline may simply be before an enemy attack or before the final damage window closes.

SPD below the next extra-turn boundary is not useless. It still moves an existing turn earlier and may change whether a buff, heal or Skill Point arrives in time.

A useful turn needs the right neighbors

In the same unmodified start, a support at 135 SPD acts at about 74.07 AV; an attacker at 134 acts at about 74.63 AV. The support can prepare an effect first, provided that effect applies to the attack. Reverse their speeds and the order reverses. This small gap proves an opening order, not a complete rotation.

If an attacker has just taken a natural turn, Bronya’s Skill can let that ally act again immediately. Used before the ally’s natural turn, it produces a different sequence. Her self-target exception still applies. Inspect the recipient, the resources spent and which turns reduce buff durations.

Check the next support action too. An opening that works can drift after temporary buffs expire, an enemy delays someone, or one teammate changes their action pattern. Team-building Fundamentals connects the timing plan to Skill Points and the team’s damage plan.

Changing SPD means tracking the remaining distance

A buff gained halfway through a turn cannot be treated as if it existed from battle start. For example, spend 40 AV at 100 SPD: 4,000 distance has passed and 6,000 remains. Raising SPD to 150 now leaves 6,000 ÷ 150 = 40 AV. The action arrives at elapsed 80 AV, not at the 66.67 AV predicted by using 150 SPD from the start.

Alternatively, at that same 6,000-distance point, apply a 25% ordinary advance while keeping 100 SPD. It removes 2,500 of the full 10,000 distance, leaving 35 AV to wait. The action arrives at elapsed 75 AV. Advance is not “25% off whatever AV is currently displayed.” Delay changes distance in the opposite direction.

Follow the timeline segment by segment whenever SPD changes or an advance occurs. Check percentage SPD buffs against base SPD, not the final stat total. Summons, extra turns, action-replacing states and special encounter rules also need their own kit conditions; the simple natural-turn table does not model them.

Choose the action before choosing the number

1. Name the action you need and its deadline.
2. Record starting distance, in-battle SPD and every effect that changes either.
3. Derive the boundary, then distinguish it from any rounded target or safety margin.
4. Check teammates, Skill Points and buff durations before and after that action.
5. Test the sequence in the encounter, including the next wave or buff expiry.

Finally, compare what the extra timing costs. Trading away a necessary main stat or too much damage for a late action can worsen the result. Use Relic Stats to evaluate the equipment trade, rather than treating SPD as the whole build.