Implicit conversion from float -5.206611570247944 to int loses precision