Implicit conversion from float -10.99693251533742 to int loses precision