Implicit conversion from float -49.60396039603961 to int loses precision