Implicit conversion from float -11.838755304101852 to int loses precision