What the model actually looks at when it makes a pick

There are about a dozen signals that move a prediction. Here's what they are and why they're there.

Feature importance explains contribution under model assumptions, not causal certainty.