Implicit conversion from float -100.69506726457399 to int loses precision