A deflection turns a comfortable save into a goal without the shooter doing anything differently. That single fact creates persistent difficulties for anyone trying to measure finishing.
What a deflection actually does
Contact with a defender changes the ball's direction late, after the goalkeeper has committed his weight and chosen his dive.
The keeper is not slow; he is beaten by information that arrived after his decision was irreversible.
The same applies to a deflection that loops rather than diverts. The trajectory the keeper spent the whole flight tracking no longer exists by the time the ball reaches him, and there is no time to rebuild it.
Shot models cannot see the touch
A model that values chances by location and situation treats a deflected shot as whatever the shot was before the touch, because that is all the shooter controlled.
That is defensible as a modelling choice, but it means the recorded value of the chance and the actual outcome of it diverge sharply on exactly the shots people remember.
Across a full league season the divergences cancel out, since deflections help and hurt roughly equally. For an individual striker taking a modest number of shots, they do not cancel at all.
Some shots invite deflections more than others
Shooting early from distance through a crowded penalty area produces more blocked shots and more deflected ones, simply because there are more bodies standing in the path of the ball.
Low, hard shots along the ground are more likely to clip a stretching leg than a rising drive struck over the same set of defenders.
So the frequency of deflections is not entirely outside a player's control, even though the outcome of any single one is. A striker who shoots through traffic has chosen a noisier route to goal.
Reputations are built on the exceptions
A deflected winner is remembered as a winner. Nobody recalls that the ball flicked off a hip on the way through.
Conversely, a striker whose clean strikes are repeatedly deflected wide gets a reputation for wastefulness he has not earned.
Judging a finisher over a handful of matches means judging the deflections as much as the finishing.
How analysts handle it
The usual approach is to flag deflected shots separately and to look at a striker's record with and without them included.
Where the two diverge substantially over a short window, the honest conclusion is that there is not yet enough evidence to draw one.
The wider the sample, the smaller the gap, which is the practical reason single-season finishing figures are treated cautiously.