Implicit conversion from float -22.156862745098053 to int loses precision