Implicit conversion from float -0.4825737265415455 to int loses precision