Implicit conversion from float -9.630996309963095 to int loses precision