Implicit conversion from float -84.19354838709677 to int loses precision