Implicit conversion from float -25.423728813559322 to int loses precision