Case A: Noisy, bad calibration R, bad calibration RMSE, but unbiased results:
Case B: Clean, good calibration R, good RMSE, but: