Implicit conversion from float -9.807692307692307 to int loses precision