Implicit conversion from float -18.366533864541836 to int loses precision