Implicit conversion from float -106.04863221884497 to int loses precision