The picture changes while an action happens
A moving target occupies a position now and may occupy another when an action reaches it. That makes timing relevant as well as direction. Before deciding how to aim, establish what the game's instructions say about the action. An immediate selection and a projectile with visible travel are different tasks; the same strategy should not be assumed to fit both.
Observe a pattern without declaring a law
Suppose a fictional target travels between two visible points. Watching it return can help you identify an interval or turning region worth considering. One repeat does not prove that the motion will always be identical. Record what you can actually distinguish: the direction of travel, the approximate region and whether a turn is approaching.
If the target changes pace or direction irregularly, predicting a distant future position may be less useful than waiting for a clearer opportunity. That is a choice about uncertainty, not evidence that the game is unfair. The movement pattern and the control system together determine which information is relevant to the decision.
Describe a candidate meeting
For an action with travel time, imagine the target and the moving object arriving at the same region. Ask whether anything blocks either route and whether the target is likely to remain in the region long enough. Avoid inventing an exact travel speed from an animation. A qualitative comparison can still distinguish an obviously late action from one worth investigating.
After an attempt, separate the intended meeting from the observed result. Did the action cross the target's path before or after it arrived? Did the target reverse direction? Did another object interrupt the route? Each observation suggests a different next question. Merely recording a miss loses information about the relationship between the paths.
Know when to stop predicting
Some games provide explicit aim assistance or instructions that change the problem. Use those explanations rather than insisting on a homemade model. If the interface shows too little to infer a reliable pattern, choose a simpler opportunity or acknowledge the uncertainty. The purpose of this method is to make timing choices understandable, not to promise a particular hit rate or claim that every moving target can be anticipated.