Implicit conversion from float -127.27272727272725 to int loses precision