Implicit conversion from float -24.503816793893122 to int loses precision