Implicit conversion from float -75.31690140845072 to int loses precision