Implicit conversion from float -29.387755102040813 to int loses precision