Implicit conversion from float -117.69230769230768 to int loses precision