Implicit conversion from float -85.73099415204678 to int loses precision