Implicit conversion from float -9.757281553398059 to int loses precision