Implicit conversion from float -22.68907563025209 to int loses precision