Implicit conversion from float -22.72727272727272 to int loses precision