Implicit conversion from float -15.78144853875476 to int loses precision