Implicit conversion from float -155.71823204419888 to int loses precision