Implicit conversion from float -9.052774018944518 to int loses precision