Implicit conversion from float -15.869872701555877 to int loses precision