Implicit conversion from float -10.492610837438434 to int loses precision