Guide

Game edge guard and ledgetrap systems explained

Harbor Brawl's platform stage had generous blast zones and a wide ledge grab box — yet only 4% of stocks ended offstage. Players knocked opponents past the stage lip, then watched them recover for free because no character tools covered low recovery arcs or punished delayed ledge options. Ground okizeme was polished; vertical stage geometry was an afterthought. The refactor added explicit edge guard coverage (hitboxes and projectiles that contest recovery before the ledge) and ledgetrap routes (pressure that limits which ledge option is safe). Offstage KO share rose to 19% without raising global knockback or shrinking the stage.

Edge guard and ledgetrap systems turn stage boundaries into authored offense. They differ from ground knockdown pressure: oki assumes a fixed knockdown point; edge play assumes a moving defender with multiple recovery vectors, ledge invulnerability windows, and stage-specific snap rules. This guide covers recovery route taxonomy, ledge option tables, edge-guard positioning, gimp tools, platform-fighter vs traditional fighting-game variants, the Harbor Brawl offstage refactor, a technique decision table versus okizeme-only tuning, pitfalls, and a production checklist.

Edge guard, ledgetrap, and gimp — three related layers

Designers often conflate offstage play into one bucket. Separating layers keeps tuning legible:

Edge guard — offense that covers recovery before the defender reaches ledge or stage. Goal: intercept high/low/sweetspot routes or force a bad position.
Ledgetrap — offense that limits ledge options once the defender is near grab range: getup attack, roll, jump, neutral rise. Goal: cover two or more options with one placement.
Gimp — low-percent tools that interrupt recovery without a full KO hitbox (spike hitboxes, weak meteor, stage spike). Goal: punish greedy routes at any damage.

A complete offstage kit usually needs all three. Characters with strong edge guard but no ledgetrap let opponents steal neutral after a safe ledge return. Characters with ledgetraps but no gimp tools feel harmless until 80% damage.

Recovery route taxonomy

Every character should document recovery as a route graph, not a single “up-B” button. Common arcs:

High route — arc over the ledge; vulnerable to anti-air and rising attacks from stage.
Low route — skims under stage lip; vulnerable to downward hitboxes, projectiles, and stage spikes.
Sweetspot / ledge-direct — aims at grab box; fastest return but predictable timing.
Deep / cross-stage — recovers to opposite side; costs resources but breaks one-sided ledgetraps.
Wall jump / air dodge — secondary mobility that must respect air action limits and invincibility rules.

Edge-guard design starts by listing which routes each character has at low, mid, and high damage — burn-knockback and hitstun change arc height. If two characters share identical recovery graphs, offstage play collapses into mirror timing tests with no read layer.

Ledge options and invulnerability windows

Once a defender reaches ledge, ledgetrap design takes over. Standard option sets:

Ledge option	Typical vulnerability	Trap counter
Neutral getup	Slow; no hitbox	Meaty low hitbox at ledge
Getup attack	Startup before active frames	Shield or spaced punish
Roll inward	Invuln then actionable	Run to cross ledge, grab
Roll away	Ends near blast zone	Chase with run or projectile
Ledge jump	Linear arc	Anti-air or rising coverage
Drop from ledge	Low recovery restart	Offstage downward hitbox

Platform fighters often add ledge invulnerability (intangible frames on grab) and two-frame punish rules (standing at ledge during grab animation). Traditional fighters may use hard knockdown at corner with different wake-up tables but the same read structure: attacker must cover multiple timings with staged hitboxes.

Ledge intangibility and fairness caps

Infinite ledge camping breaks pacing. Common caps: limited re-grabs without touching stage, gradual grab box shrink, or % -based ledge slip. Document these caps beside trap tables so balance patches do not accidentally remove a character's only recovery route.

Positioning: stage control vs offstage commitment

Edge guard is a risk contract. The attacker leaves stage to contest recovery; a whiffed deep jump often costs a stock to reverse edge guard. Good systems make commitment readable:

Recovery point — how many air actions remain when reaching ledge; designers should surface this in training UI.
Ledge coverage zones — annotate low / mid / high threat radii from the ledge lip in editor tools.
Return paths — after offstage jump, how does attacker recover? Wall jump, up-special, or death if missed.
Projectile edge guard — lower risk but slower kill confirm; must respect projectile clank and lifetime at blast zone depth.

Stage layout matters as much as character kits. Narrow underside geometry rewards low coverage; wide blast zones reward patient ledgetrap over early deep commits. Tie blast zone distance to average kill percent so edge guard remains a mid-match layer, not only a 120% swing.

Harbor Brawl offstage refactor

Harbor Brawl's initial offstage kit was three universal moves: a rising attack, a horizontal projectile, and a single meteor on one character. Telemetry showed defenders always took high route because low route had zero threat. Ledgetrap wins were accidental grab punishes, not authored traps.

The refactor shipped in four passes:

Route threat map — every character gained at least one low-coverage tool (down-tilt off ledge, arcing projectile, or lingering hitbox) with documented active frames at blast-zone depth.
Ledge option table per character — two-option traps (e.g., roll-in meaty + jump anti-air) instead of one universal grab check.
Gimp tier — weak meteor at 6–9% damage on four characters; spikes at high knockback unchanged.
Training overlay — recovery route ghosts and ledge trap success rate, mirroring the oki training mode from the wake-up refactor.

Results after two balance patches: offstage KO share 4% → 19%; average match length unchanged; rematch rate +6% (players reported offstage reads as “fair mindgames” in surveys). Deep offstage suicide rate for attackers rose 2pp — acceptable trade for interactive recovery.

Technique decision table

Goal	Prefer	Avoid
Kill at high % offstage	Edge guard with strong horizontal knockback at blast zone	Ground oki strings only — no route coverage
Punish predictable recovery	Ledgetrap covering roll + neutral getup	Chasing with full committal deep hits every time
Low-% route conditioning	Gimp tools and stage spikes	Raising global knockback to force early KOs
Accessible offstage play	Two-option traps with training overlay feedback	Frame-perfect 1f ledge grabs with no counterplay
Traditional fighter corner	Carry + oki layered on wall bounce	Platform ledge rules copied without wake-up tables

Common pitfalls

One recovery route dominant — if high route is always safest, edge guard devolves into single anti-air.
Ledge invuln not documented — players blame “broken grab” when intangibility frames eat meaties.
Offstage kills only at extreme % — edge layer never activates in mid-game.
No attacker recovery — deep edge guard unused because whiff equals death; players stay on stage.
Identical ledge traps on every stage — ignores underside geometry and blast zone shape.
Gimp without kill confirm — annoyance without payoff; players feel stalled not threatened.

Production checklist

Document recovery route graph per character at low / mid / high knockback.
Publish ledge option table with startup, invuln, and active frames.
Ensure each character has low, mid, and high edge-guard coverage tools.
Author at least two two-option ledgetraps per archetype (rushdown, zoner, grappler).
Cap ledge re-grab or document slip rules to prevent infinite camping.
Annotate blast zones and ledge grab boxes in stage editor.
Build training mode overlay for route coverage and trap success rate.
Test offstage KO share target (typically 12–25% for platform fighters).
Validate attacker recovery paths after deep edge commits.
Cross-check with corner escape systems so ground trap and offstage trap do not contradict.

Key takeaways

Edge guard, ledgetrap, and gimp are separate authored layers.
Recovery is a route graph — coverage must threaten multiple arcs.
Ledge play is option coverage, structurally similar to okizeme but in vertical space.
Offstage commitment needs attacker recovery paths or players will not engage.
Harbor Brawl raised offstage KO share from 4% to 19% with route maps and two-option traps.