Implicit conversion from float -91.13924050632912 to int loses precision