Implicit conversion from float -120.66176470588238 to int loses precision