Implicit conversion from float -11.95804195804196 to int loses precision