Implicit conversion from float -31.182266009852228 to int loses precision