Implicit conversion from float -14.842931937172764 to int loses precision