Implicit conversion from float -14.70779220779221 to int loses precision