Implicit conversion from float -19.16030534351144 to int loses precision