Implicit conversion from float -10.26420079260238 to int loses precision