Implicit conversion from float -166.15384615384613 to int loses precision