Implicit conversion from float -14.697986577181211 to int loses precision