Implicit conversion from float -18.099173553719012 to int loses precision