Implicit conversion from float -11.522842639593904 to int loses precision