Implicit conversion from float -41.53846153846155 to int loses precision