Implicit conversion from float -52.13483146067415 to int loses precision