Implicit conversion from float -12.966101694915267 to int loses precision