Implicit conversion from float -9.855282199710572 to int loses precision