Risk Model Drift & Retraining Monitor — 2D

AI credit and fraud risk models don't stay accurate forever — as the economy shifts, the population of transactions and applicants a model scores drifts away from the data it was trained on, and its predictions quietly degrade. This 2D edition renders that process as a ridge (joyplot) of daily score-distribution histograms scrolling back through simulated time, drifting away from a frozen training baseline as you tune how fast and how hard the market regime shifts. Drag to pan across the ridge and scroll or pinch to zoom in on any window of days. A live Population Stability Index (PSI) — the exact statistic banks use to monitor deployed risk models under Basel-III-era governance — tracks how far the two distributions have diverged, model accuracy degrades as drift grows, and you can trigger a volatility shock or retrain the model on demand to watch the whole cycle reset.