Trim TimesFM backcast from forecasts

models/timesfm_model.py

@@ -81,7 +81,10 @@ class TimesFMModel(BaseForecastModel):
         model = self._ensure_compiled_for(len(context))
         if features is None:
             point_forecast, _quantiles = model.forecast(horizon=horizon, inputs=[context])
-            point_values = np.asarray(point_forecast[0], dtype=float)
+            # TimesFM 2.5 returns the requested forecast after an optional
+            # backcast when ``return_backcast`` is enabled.  The backcast is
+            # useful to XReg but is not part of this model's forecast API.
+            point_values = np.asarray(point_forecast[0], dtype=float)[-horizon:]
         else:
             feat_context = features.iloc[-len(context):].reset_index(drop=True)
             validate_features(feat_context, expected_len=len(context), model_name='TimesFM')
@@ -95,7 +98,7 @@ class TimesFMModel(BaseForecastModel):
                 dynamic_numerical_covariates[col] = [full_array]
             try:
                 point_forecast, _quantiles = model.forecast_with_covariates(inputs=[context[1:]], dynamic_numerical_covariates=dynamic_numerical_covariates, xreg_mode='xreg + timesfm')
-                point_values = np.asarray(point_forecast[0], dtype=float)
+                point_values = np.asarray(point_forecast[0], dtype=float)[-horizon:]
             except (ImportError, AttributeError) as e:
                 raise ImportError(f"TimesFM's covariate forecasting needs the 'xreg' extra (scikit-learn + JAX). Install with:\n  pip install timesfm[torch,xreg]\n(original error: {e})")
         if point_values.shape != (horizon,):
@@ -103,5 +106,5 @@ class TimesFMModel(BaseForecastModel):
         out = point_values.tolist()
         if not all(np.isfinite(x) and x > 0 for x in out):
             raise ValueError('TimesFM returned a non-finite forecast (NaN/inf) for this window. This is usually a sign the input window is degenerate (e.g. a flat or near-constant price run) — try a different window size or symbol.')
-        self.last_diagnostics = {'checkpoint': self.checkpoint, 'context_candles': len(context), 'feature_policy': 'lagged xreg; unknown future covariates persist', 'fallback': False}
+        self.last_diagnostics = {'checkpoint': self.checkpoint, 'context_candles': len(context), 'feature_policy': 'lagged xreg; unknown future covariates persist', 'backcast_trimmed': True, 'fallback': False}
         return out