What if the RBI moved the repo rate today?
Drag the slider below to simulate a hypothetical repo-rate change. The trained XGBoost model re-runs over the last 12 observed weeks, perturbing the repo rate and its downstream lagged and spread features. The shaded band is the 90% confidence interval derived from walk-forward residuals.
How to read this. Because WACMR tracks the RBI rate corridor tightly, a +/-25 bps move often shows a small net effect here — the model's top feature is last week's WACMR (target_lag1, mean |SHAP| ≈ 0.49), which does not change under the counterfactual. Larger moves (±100 bps, ±200 bps) show the rate-corridor channel more clearly. The asymmetry between hikes and cuts reflects the tree model's learned response across the two monetary-policy regimes.
This is a model-basedcounterfactual, not a causal one. It answers "how would the XGBoost forecast respond?", not "what would actually happen in the economy?".