Implicit conversion from float -10.19083969465649 to int loses precision