Implicit conversion from float -24.32926829268294 to int loses precision