Implicit conversion from float -9.409594095940975 to int loses precision