Implicit conversion from float -51.75824175824175 to int loses precision