Tracking is the ability of a move to hit a sidestepping opponent. It's affected by the move:
- having a wider hitbox
- having a hitbox closer to one side
- realigning towards the opponent during startup
In addition, there are moves which cause either the attacker or defender to recover slightly off-axis, which can change the tracking of immediate followups.
Measurement
The following tests are used to measure tracking scores for Wavu Wiki movelists:
[...TABLE...]
The tracking score is equal to startup - f, where startup is the startup of the move, and f is the test with the worst frame advantage where the move still hits. However, if it misses in a test with better frame advantage, then the worse test doesn't count. This is to avoid weird situations where sometimes a move will hit at e.g. -9 but not -2 to -8.
The tracking score is thus the number of frames the opponent can sidestep for and still be hit by the move.
In tests where the bot has around 11 to 14 frames to sidestep, both sidestep and sidewalk should be tested. The test is passed only if both get hit.
Considering startup
An often overlooked aspect to measuring tracking is that a move's speed plays a role. Slower moves give the opponent more time to sidestep, which leads to two unexpected results:
- If tests measure if a move hits after a sidewalk at +0, then slow moves will be underrated.
- If tests measure if a move hits after a sidestep, then slow moves will be overrated.
For this reason, the âtracking scoreâ takes startup into account.
Knowing this is necessary to contextualize the score: Against an opponent doing sidestep block, a slow move that tracks 15 frames of sidestep is just as good as a fast one. But against an opponent who commits to a sidewalk, the faster move is better, because it doesn't give them that extra time.
Limitations
This measurement is simplifying things.
In some cases, a move will track well at close range but not at long range, or it can hit at -9 but not at -5, because the hitbox has a weird shape. In addition, there are myriad interactions between odd hurtboxes which aren't reflected by testing only against a single, standardized opponent.
There are also many situations where a move will hit an opponent sidestepping but whiff against sidestep guard or sidestep crouch, and vice versa.
The most significant limitations is that the tests only work on moves done from standing. Moves from stances must be tested in other ways.
Alternatives
One could simply note the exact size and position of the hitbox and whether or not it realigns, but for most players this is not useful information.
One could reduce the score down to something like âtracks leftâ, âtracks rightâ, or âno trackingâ, but this would be oversimplifying.