Implicit conversion from float -24.663461538461533 to int loses precision